CN109903224B - Image scaling method and device, computer equipment and storage medium - Google Patents

Abstract

The application relates to an image scaling method, an image scaling device, a computer device and a storage medium. The method comprises the following steps: acquiring pixel values of original pixel points in a preset neighborhood around a mapping coordinate point of a pixel point to be interpolated in a target image; determining edge gradient direction information and edge strength information of the mapping coordinate points according to pixel values of the original pixel points; determining the weight of each original pixel point in the neighborhood; determining an edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, the edge strength information and the weight; weighting the weight of each original pixel point by using the edge enhancement attenuation correction coefficient to obtain a convolution filtering weight; and determining the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filtering weight and the pixel value of each original pixel point in the neighborhood. By adopting the method, the edge information of the original image can be kept in the zooming process, and the quality of the zoomed target image is improved.

Description

Image scaling method, device, computer equipment and storage medium

Technical Field

The present application relates to the field of image processing technology, and in particular to an image scaling method, device, computer equipment and storage medium.

Background Art

With the development of display technology, the demand for image scaling technology between different resolutions is also increasing. For example, in high-definition or ultra-high-definition display technology, image scaling technology is needed to scale existing video resources to a matching size so as to be scaled and displayed on a small screen or ultra-high-definition screen to meet the display resolution requirements.

Traditional image scaling methods mainly include low-pass filtering-based interpolation methods and edge-based interpolation methods. Among them, low-pass filtering interpolation methods include bilinear interpolation, bicubic interpolation, etc. In traditional image interpolation algorithms, neighbor interpolation is simpler and easier to implement. It was widely used in the early days, but this method will produce obvious jagged edges and mosaic phenomena in the new image. Bilinear interpolation has a smoothing function and can effectively overcome the shortcomings of the neighbor method, but it will degenerate the high-frequency part of the image and blur the image details. When the magnification is relatively high, high-order interpolation, such as bicubic and cubic spline interpolation, is better than low-order interpolation.

Traditional image scaling methods use interpolation operations to make the interpolated pixel grayscale values continue the continuity of the original image grayscale changes, so that the image obtained after scaling has natural and smooth changes in light and dark. However, in the image, there are mutations in the grayscale values between some pixels and adjacent pixels, that is, there is grayscale discontinuity. These pixels with grayscale mutations are the edge pixels of the image that describe the outline of the object or the texture image. In image magnification, for these pixels with discontinuous grayscale characteristics, if conventional interpolation algorithms are used to generate new pixels, the outline and texture of the enlarged image will inevitably be blurred, and the edge information will be lost, resulting in poor image quality after scaling.

Summary of the invention

Based on this, it is necessary to provide an image scaling method, device, computer equipment and storage medium that can retain image edge information and improve the quality of the target image obtained after scaling to address the above technical problems.

An image scaling method, the method comprising:

Obtain the pixel point to be interpolated in the target image; determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image; determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood; determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood; determine the edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance; weight the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; determine the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

An image scaling device, comprising:

The module for obtaining pixels to be interpolated is used to obtain pixels to be interpolated in the target image;

A neighborhood pixel acquisition module is used to determine the mapping coordinate point of the pixel point to be interpolated in the original image, and to obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image;

An edge information determination module is used to determine edge gradient direction information and edge intensity information of a mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

The original weight determination module is used to determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapped coordinate point and each original pixel point in the neighborhood;

The enhancement attenuation correction coefficient determination module is used to determine the edge enhancement attenuation correction coefficient of each original pixel in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance;

The pixel value determination module for the pixel point to be interpolated is used to weight the weights of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood, the pixel value of the pixel point to be interpolated in the target image is determined.

A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

Obtain the pixel point to be interpolated in the target image; determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image; determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood; determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood; determine the edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance; weight the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; determine the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

Obtain the pixel point to be interpolated in the target image; determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image; determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood; determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood; determine the edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance; weight the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; determine the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

The above-mentioned image scaling method, device, computer equipment and storage medium determine the edge gradient direction information and edge strength information of the mapping coordinate point according to the pixel values of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image corresponding to the pixel point to be interpolated, enhance or attenuate the weight of each original pixel point in the neighborhood to obtain the weight of each original pixel point after enhancement or attenuation, and solve the pixel value of the pixel point to be interpolated according to the weight of each original pixel point after enhancement or attenuation and the pixel value of each original pixel point. It is able to maintain the correlation between the pixels before and after during the interpolation process, and can perform interpolation processing on the edges in any direction to maintain the edge information of the original image, so that the image remains clear after being enlarged or reduced, avoiding the occurrence of jagged distortion and other phenomena, retaining the integrity and correlation of the image edge information, making the contour and texture of the scaled image clearer, and improving the quality of the target image obtained after scaling.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of a flow chart of an image scaling method in one embodiment;

FIG2 is a schematic diagram of a flow chart of an image scaling method in another embodiment;

FIG3 is a schematic diagram of an embodiment;

FIG4 is a schematic diagram of an embodiment;

FIG5 is a schematic diagram of an embodiment;

FIG6 is a schematic diagram of an embodiment;

FIG7 is a schematic diagram of an embodiment;

FIG8 is a structural block diagram of an image scaling device in one embodiment;

FIG. 9 is a diagram showing the internal structure of a computer device in one embodiment.

DETAILED DESCRIPTION

In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

The image scaling method provided in the present application can be applied to a terminal, wherein the terminal can be, but is not limited to, various televisions, personal computers, notebook computers, smart phones, tablet computers, portable wearable devices, and the like.

In one embodiment, as shown in FIG1 , an image scaling method is provided, comprising the following steps S110 - S180:

S110, obtaining pixels to be interpolated in the target image;

Among them, the target image is the image obtained after scaling the original image, the image difference is the process of calculating the unknown pixels in the target image using the known pixels in the original image, and the pixels to be interpolated refer to the unknown pixels in the scaled target image that need to be determined by calculation.

S120, determining the mapping coordinate point of the pixel point to be interpolated in the original image;

As an example, as shown in FIG3 , it is assumed that the input original image is enlarged to a high-resolution target image with a low resolution, and the width and height of the original image are W _src and h _src , respectively, and the width and height of the target image are W _tar and h _{tar ,} respectively. Then the coordinate value (I _tar , J _tar ) of the pixel point to be interpolated in row I and column J of the target image is the coordinate value of the mapping coordinate point in the original image (I _src , J _src ):

S130, obtaining the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image;

Among them, the original pixel points in the preset neighborhood around the mapping coordinate point can determine the number of original pixel points in the neighborhood according to actual needs, for example, it can be 2×2, 3×4, 3×3 or 6×6 original pixel points around the original pixel point. The more original pixel points in the neighborhood are used, the better the target image effect generated after scaling and the more complete the information, but the amount of calculation will increase accordingly, so the appropriate number of original pixel points in the neighborhood can be selected according to the scaling situation.

As an example, as shown in FIG4 , the original pixels in a 4×4 neighborhood around the coordinate point (I _src , J _src ) mapped to the pixel point to be interpolated (I _tar , J _tar ) in the original image may be selected as neighborhood pixels.

S140, determining edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

Among them, the pixel value is a quantity that characterizes the value of a pixel point. For example, it can be the channel value of a pixel point in a certain color space. Taking the RGB color space as an example, the pixel value of a pixel point can be expressed as (R, G, B), where R, G, and B are the channel values of the three channels of the RGB color space respectively.

In one embodiment, as shown in FIG. 2 , S140 determines the edge gradient direction information and edge strength information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood, including S141-S142:

S141, determining a horizontal edge gradient value and a corresponding horizontal edge intensity value of a mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

In one embodiment, S141 determines the horizontal edge gradient value and the corresponding horizontal edge strength value of the mapping coordinate point according to the pixel values of each original pixel point in the neighborhood, including: calculating the average of the pixel values of each original pixel point in the neighborhood; obtaining the original pixel points of a preset row among the original pixel points in the neighborhood; determining the horizontal edge gradient value of the mapping coordinate point according to the difference in the horizontal direction of the original pixel points of the preset row; determining the horizontal edge strength value of the mapping coordinate point according to the difference between the original pixel points of the preset row and the average value.

Further, in one embodiment, obtaining original pixel points of a preset row among original pixel points in a neighborhood includes: obtaining original pixel points of a number of rows other than the first row and the last row among original pixel points in the neighborhood as original pixel points of the preset row.

As an example, taking the original pixel points in a 4×4 neighborhood around the mapping coordinate point (I _src , J _src ) as an example, the mean value of the pixel values of each original pixel point in the neighborhood is calculated as follows:

In the above formula, avg is the average pixel value of each original pixel in the neighborhood, i is the row subscript of the original pixel in the neighborhood, i takes a value of 0 to 3, j is the column subscript of the original pixel in the neighborhood, j takes a value of 0 to 3, and pi _,j is the pixel value of the original pixel (i, j) in the neighborhood.

The horizontal edge gradient value of the mapping coordinate point (I _src , J _src ) is calculated as follows:

h_diff0＝-1*p ₁₀ -2*p ₁₁ +2*p ₁₂ +1*p ₁₃

h_diff1＝-1*p ₂₀ -2*p ₂₁ +2*p ₂₂ +1*p ₂₃

h_diff＝h_diff0+h_diff1

In the above formula, h_diff0 is the horizontal difference of the original pixel points in the first row in the neighborhood, h_diff1 is the horizontal difference of the original pixel points in the second row in the neighborhood, and h_diff is the horizontal edge gradient value of the mapping coordinate point.

The horizontal edge strength value corresponding to the mapping coordinate point (I _src , J _src ) is calculated as follows:

In the above formula, h_var is the horizontal edge strength value, and abs is the absolute value.

In the above embodiment, in order to simply calculate the horizontal edge gradient value and the corresponding horizontal edge intensity value of the original pixel point in the 4×4 neighborhood of the mapping coordinate point, the first row and the second row of pixel values with greater correlation among the original pixel points in the neighborhood are used to calculate the horizontal edge gradient value and the horizontal edge intensity value (the original pixel points in the neighborhood are 0, 1, 2, 3 rows, and 0, 1, 2, 3 columns). Since the first column and the second column of each row in the neighborhood have a higher correlation with the interpolated pixel point than the zeroth column and the third column, the matrix of [-1, -2, 2, 1] ^T is used to calculate the horizontal difference of the original pixel points in the first row and the second row, and the horizontal gradient direction of the edge is determined according to the positive or negative sum of the row direction difference. The corresponding difference is obtained according to the calculated pixel average value in the neighborhood and the pixel values of each column in the first row and the second row, and the absolute value is summed to obtain the horizontal edge intensity value as the horizontal direction intensity information for determining the edge.

S142, determining the vertical edge gradient value and the corresponding vertical edge intensity value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood.

In one embodiment, S142 determines the vertical edge gradient value and the corresponding vertical edge strength value of the mapping coordinate point according to the pixel values of each original pixel point in the neighborhood, including: calculating the average of the pixel values of each original pixel point in the neighborhood; obtaining the original pixel points of a preset column among the original pixel points in the neighborhood; determining the vertical edge gradient value of the mapping coordinate point according to the difference in the vertical direction between the original pixel points of the preset column; determining the vertical edge strength value of the mapping coordinate point according to the difference between the original pixel points of the preset column and the average value.

In one embodiment, obtaining original pixel points of a preset column among original pixel points in a neighborhood includes: obtaining original pixel points of a number of columns other than the first column and the last column among original pixel points in the neighborhood as original pixel points of the preset column.

As an example, also taking the original pixel points in the 4×4 neighborhood around the mapping coordinate point (I _src , J _src ) as an example, the mean value of the pixel values of each original pixel point in the neighborhood is calculated as follows:

In the above formula, avg is the average pixel value of each original pixel in the neighborhood, i is the row subscript of the original pixel in the neighborhood, i takes a value of 0 to 3, j is the column subscript of the original pixel in the neighborhood, j takes a value of 0 to 3, and pi _,j is the pixel value of the original pixel (i, j) in the neighborhood.

The vertical edge gradient value of the mapping coordinate point (I _src , J _src ) is calculated as follows:

v_diff0＝-1*p ₀₁ -2*p ₁₁ +2*p ₂₁ +1*p ₃₁

v_diff1＝-1*p ₀₂ -2*p ₁₂ +2*p ₂₂ +1*p ₃₂

v_diff＝v_diff0+v_diff1

In the above formula, v_diff0 is the vertical difference of the original pixel points in the first row in the neighborhood, v_diff1 is the vertical difference of the original pixel points in the second row in the neighborhood, and v_diff is the vertical edge gradient value of the mapping coordinate point.

The vertical edge strength value corresponding to the mapping coordinate point (I _src , J _src ) is calculated as follows:

In the above formula, v_var is the vertical edge strength value, and abs is the absolute value.

In the above embodiment, in order to simply calculate the vertical edge gradient value and the corresponding vertical edge strength value of the original pixel point in the 4×4 neighborhood of the mapping coordinate point, the first and second column pixel values with greater correlation in the original pixel points in the neighborhood are used to calculate the vertical edge gradient value and the vertical edge strength value (the original pixel points in the neighborhood are 0, 1, 2, 3 rows, and 0, 1, 2, 3 columns). Since the first and second rows of each column in the neighborhood have a higher correlation with the interpolated pixel points than the zeroth row and the third row, the matrix of [-1 -2 2 1] is used to calculate the difference in the vertical direction of the original pixel points in the first and second columns, and the vertical gradient direction of the edge is determined according to the positive or negative sum of the column direction difference. The corresponding difference is obtained based on the calculated pixel average value in the neighborhood and the pixel values of each row in the first and second columns, and the absolute value is summed to obtain the vertical edge strength value as the vertical direction strength information for determining the edge.

S150, determining the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapped coordinate point and each original pixel point in the neighborhood;

As an example, the Gaussian distance between the mapped coordinate point and each original pixel point in the neighborhood is calculated as follows:

Where (x, y) is the coordinate value of any original pixel point in a 4*4 neighborhood centered on the pixel point to be interpolated, and r _sd (x, y) is the Gaussian distance between the mapped coordinate point and the original pixel point with coordinates (x, y) in the neighborhood;

Taking the original pixel points in a 4×4 neighborhood as neighborhood pixels as an example, in this step, 16 original pixel points correspond to 16 coordinate values (x, y), and 16 Gaussian distances r _sd (x, y) can be calculated respectively.

The weight of the original pixel in the corresponding neighborhood is determined according to r _sd (x, y) as shown in the following formula:

Where (x, y) is the coordinate value of the original pixel point in the neighborhood, w _{(x, y)} is the weight corresponding to r _sd (x, y), n _x is the normalization coefficient, r _sd (x, y) is the distance between the original pixel point in the neighborhood and the mapped coordinate point, and σ is the standard deviation of the corresponding Gaussian distribution;

Taking the original pixels in a 4×4 neighborhood as neighborhood pixels as an example, in this step, 16 weights of the 16 original pixels can be calculated according to the 16 Gaussian distances r _sd (x, y).

The above is the basic principle of weight calculation in this step. In practical applications, in order to simplify the calculation, in this step, the weights corresponding to different Gaussian distances can be directly obtained by querying the stored weight table. Specifically, the corresponding weight value can be obtained by looking up the table according to the Gaussian distance of each point in the neighborhood based on the corresponding smooth low-pass filtered Gaussian curve. The normalization coefficient _nx corresponds to the inverse of the sum of the weights of each point in the neighborhood.

In this embodiment, the Gaussian distance can effectively derive the contribution weight of each original pixel in the neighborhood to the current pixel to be interpolated, thereby retaining the correlation between the pixel to be interpolated and the original pixel in the neighborhood of the original image to a certain extent, so that the image edge features in the original image are retained to a certain extent.

S160, determining an edge enhancement attenuation correction coefficient of each original pixel in the neighborhood according to the edge gradient direction information, the edge intensity information and the Gaussian distance;

In one embodiment, as shown in FIG. 2 , S160 determines the edge enhancement attenuation correction coefficient of each original pixel in the neighborhood according to the edge gradient direction information, the edge intensity information and the Gaussian distance, including S161-S163:

S161, determining a horizontal edge correction coefficient of each original pixel point according to a horizontal edge gradient value, a horizontal edge intensity value, and a Gaussian distance of each original pixel point relative to the mapped coordinate point;

In one embodiment, S161 determines the horizontal edge correction coefficient of each original pixel point according to the horizontal edge gradient value, the horizontal edge intensity value and the Gaussian distance of each original pixel point relative to the mapping coordinate point, including: determining the positive or negative nature of the horizontal edge enhancement attenuation exponent according to the horizontal edge gradient value of the mapping coordinate point; determining the size of the horizontal edge enhancement attenuation base according to the horizontal edge intensity value of the mapping coordinate point; for each original pixel point, using the horizontal edge enhancement attenuation base as the base and the product of the horizontal edge enhancement attenuation exponent and the Gaussian distance from the original pixel point to the mapping coordinate point as the exponent, performing a power operation to obtain the horizontal edge correction coefficient of the original pixel point.

As an example, according to the horizontal edge gradient value of the mapping coordinate point, the positive and negative of the horizontal edge enhancement attenuation index is determined as shown in the following formula:

In the above formula, h_signed is the horizontal edge enhancement attenuation index.

According to the horizontal edge strength value of the mapping coordinate point, the size of the horizontal edge enhancement attenuation base α is determined;

For each original pixel point, the horizontal edge enhancement attenuation base is used as the base, and the product of the horizontal edge enhancement attenuation exponent and the Gaussian distance from the original pixel point to the mapping coordinate point is used as the exponent. The horizontal edge correction coefficient of the original pixel point is obtained by performing a power operation as shown in the following formula:

In the above formula, α′(x,y) is the horizontal edge correction coefficient of the original pixel with coordinates (x,y).

The above embodiment is the horizontal edge correction coefficient calculated according to the principle in this step S161. In practical applications, in order to simplify the calculation, in one embodiment, S161 determines the horizontal edge correction coefficient of each original pixel point according to the horizontal edge gradient value, the horizontal edge intensity value and the Gaussian distance of each original pixel point relative to the mapping coordinate point of the mapping coordinate point, including: determining a corresponding curve set according to the positive and negative nature of the horizontal edge gradient value of the mapping coordinate point; the curve set includes an enhancement curve set with a curve slope greater than zero or an attenuation curve set with a curve slope less than zero; selecting a corresponding curve from the determined curve set according to the horizontal edge intensity value, and the absolute value of the slope of the horizontal edge intensity value and the selected corresponding curve is positively proportional; querying the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate from the selected curve as the horizontal edge correction coefficient of each original pixel point.

As an example, when the horizontal edge gradient value h_diff of the mapping coordinate point is greater than or equal to zero, it is determined as an enhancement curve set; wherein the enhancement curve set includes a preset number of enhancement curves, each enhancement curve corresponding to a horizontal edge gradient value h_diff in a value interval;

As shown in FIG. 5 , it is a schematic diagram of 7 enhancement curves in an enhancement curve set in one embodiment. According to the horizontal edge intensity value h_var, the corresponding enhancement curve is selected from the enhancement curve set as follows:

When h_var＜strength_level_th0, select level0 curve;

When strength_level_th0≤h_var＜strength_level_th1, select level1 curve;

When strength_level_th1≤h_var＜strength_level_th2, select level2 curve;

When strength_level_th2≤h_var＜strength_level_th3, select level3 curve;

When strength_level_th3≤h_var＜strength_level_th4, select level4 curve;

When strength_level_th4≤h_var＜strength_level_th5, select level5 curve;

When strength_level_th5≤h_var＜strength_level_th6, select level6 curve;

When strength_level_th6≤h_var＜strength_level_th7, select level7 curve;

When h_var≥strength_level_th7, select level8 curve;

Among them, strength_level_th0~7 represent the threshold values of the preset value range, which can be set according to the needs, and the slopes of the corresponding curves level1~8 increase successively; the slope of the level0 curve is 0, that is, no enhancement is performed, and the corresponding α′(x,y) values queried in the curve for each original pixel point (x,y) are all 1.

After the corresponding enhancement curve is selected, the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate is queried from the selected enhancement curve as the horizontal edge correction coefficient of each original pixel point.

When the horizontal edge gradient value h_diff of the mapping coordinate point is less than zero, it is determined as an attenuation curve set; wherein the attenuation curve set includes a preset number of attenuation curves, and each attenuation curve corresponds to a horizontal edge gradient value h_diff in a value interval;

As shown in FIG6 , it is a schematic diagram of 7 attenuation curves in an attenuation curve set in one embodiment. According to the horizontal edge intensity value h_var, the corresponding attenuation curve is selected from the attenuation curve set as follows:

When h_var＜strength_level_th0, select level0 curve;

When strength_level_th0≤h_var＜strength_level_th1, select level1 curve;

When strength_level_th1≤h_var＜strength_level_th2, select level2 curve;

When strength_level_th2≤h_var＜strength_level_th3, select level3 curve;

When strength_level_th3≤h_var＜strength_level_th4, select level4 curve;

When strength_level_th4≤h_var＜strength_level_th5, select level5 curve;

When strength_level_th5≤h_var＜strength_level_th6, select level6 curve;

When strength_level_th6≤h_var＜strength_level_th7, select level7 curve;

When h_var≥strength_level_th7, select level8 curve;

Among them, strength_level_th0~7 represent the threshold values of the preset value range, which can be set according to the needs. The slopes of the corresponding curves level1~8 decrease successively (the absolute values of the slopes increase successively), among which the slope of the level0 curve is 0, that is, no attenuation is performed, and the corresponding α′(x,y) values queried in the curve for each original pixel point (x,y) are all 1.

After the corresponding attenuation curve is selected, the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapped coordinate point as the horizontal coordinate is queried from the selected attenuation curve as the horizontal edge correction coefficient of each original pixel point.

In other embodiments, there are multiple alternative solutions to the above-mentioned solution of obtaining the corresponding horizontal edge correction coefficient through curve query. For example, the above-mentioned curve can be converted into a corresponding table, and the horizontal edge correction coefficient of each original pixel point can be queried and determined by table lookup based on the horizontal edge gradient value, horizontal edge intensity value and Gaussian distance of each original pixel point relative to the mapping coordinate point of the mapping coordinate point, and the like.

S162, determining a vertical edge correction coefficient of each original pixel point according to a vertical edge gradient value, a vertical edge intensity value, and a Gaussian distance of each original pixel point relative to the mapped coordinate point;

In one embodiment, S162 determines the vertical edge correction coefficient of each original pixel point according to the vertical edge gradient value, the vertical edge strength value and the Gaussian distance of each original pixel point relative to the mapping coordinate point of the mapping coordinate point, including: determining a corresponding curve set according to the positive and negative vertical edge gradient value of the mapping coordinate point; the curve set includes an enhanced curve set with a curve slope greater than zero or an attenuation curve set with a curve slope less than zero; selecting a corresponding curve from the determined curve set according to the vertical edge strength value, and the vertical edge strength value is positively proportional to the absolute value of the slope of the selected corresponding curve; querying the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate from the selected curve as the vertical edge correction coefficient of each original pixel point.

As an example, according to the vertical edge gradient value of the mapping coordinate point, the positive and negative of the vertical edge enhancement attenuation index is determined as shown in the following formula:

In the above formula, v_signed is the vertical edge enhancement attenuation index.

According to the vertical edge strength value of the mapping coordinate point, the size of the vertical edge enhancement attenuation base α is determined;

For each original pixel point, the vertical edge enhancement attenuation base is used as the base, and the vertical edge enhancement attenuation exponent and the product of the Gaussian distance from the original pixel point to the mapping coordinate point are used as the exponent. The vertical edge correction coefficient of the original pixel point is obtained by performing a power operation as shown in the following formula:

In the above formula, β′(x, y) is the vertical edge correction coefficient of the original pixel with coordinates (x, y).

The above embodiment is the vertical edge correction coefficient calculated according to the principle in this step S162. In practical applications, in order to simplify the calculation, in one embodiment, S162 determines the vertical edge correction coefficient of each original pixel point according to the vertical edge gradient value, the vertical edge intensity value and the Gaussian distance of each original pixel point relative to the mapping coordinate point of the mapping coordinate point, including: determining a corresponding curve set according to the positive and negative vertical edge gradient value of the mapping coordinate point; the curve set includes an enhancement curve set with a curve slope greater than zero or an attenuation curve set with a curve slope less than zero; selecting a corresponding curve from the determined curve set according to the vertical edge intensity value, and the vertical edge intensity value is proportional to the absolute value of the slope of the selected corresponding curve; querying the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate from the selected curve as the vertical edge correction coefficient of each original pixel point.

As an example, when the vertical edge gradient value v_diff of the mapping coordinate point is greater than or equal to zero, it is determined as an enhancement curve set; wherein the enhancement curve set includes a preset number of enhancement curves, each enhancement curve corresponding to a vertical edge gradient value v_diff in a value interval;

As shown in FIG. 5 , it is a schematic diagram of 7 enhancement curves in an enhancement curve set in one embodiment. According to the vertical edge strength value v_var, the corresponding enhancement curve is selected from the enhancement curve set as follows:

When v_var＜strength_level_th0, select level0 curve;

When strength_level_th0≤v_var＜strength_level_th1, select level1 curve;

When strength_level_th1≤v_var＜strength_level_th2, select level2 curve;

When strength_level_th2≤v_var＜strength_level_th3, select level3 curve;

When strength_level_th3≤v_var＜strength_level_th4, select level4 curve;

When strength_level_th4≤v_var＜strength_level_th5, select level5 curve;

When strength_level_th5≤v_var＜strength_level_th6, select level6 curve;

When strength_level_th6≤v_var＜strength_level_th7, select level7 curve;

When v_var≥strength_level_th7, select level8 curve;

Among them, strength_level_th0~7 represent the thresholds of the preset value range, which can be set according to needs. The slopes of the corresponding curves level1~8 increase successively, and the slope of the level0 curve is 0, that is, no enhancement is performed, and the corresponding β′(x,y) values queried in the curve for each original pixel point (x, y) are all 1.

After the corresponding enhancement curve is selected, the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapped coordinate point as the horizontal coordinate is queried from the selected enhancement curve as the vertical edge correction coefficient of each original pixel point.

When the vertical edge gradient value v_diff of the mapping coordinate point is less than zero, it is determined to be an attenuation curve set; wherein the attenuation curve set includes a preset number of attenuation curves, and each attenuation curve corresponds to a vertical edge gradient value v_diff in a value interval;

As shown in FIG. 6 , it is a schematic diagram of 7 attenuation curves in an attenuation curve set in one embodiment. According to the vertical edge strength value v_var, the corresponding attenuation curve is selected from the attenuation curve set as follows:

When v_var＜strength_level_th0, select level0 curve;

When strength_level_th0≤v_var＜strength_level_th1, select level1 curve;

When strength_level_th1≤v_var＜strength_level_th2, select level2 curve;

When strength_level_th2≤v_var＜strength_level_th3, select level3 curve;

When strength_level_th3≤v_var＜strength_level_th4, select level4 curve;

When strength_level_th4≤v_var＜strength_level_th5, select level5 curve;

When strength_level_th5≤v_var＜strength_level_th6, select level6 curve;

When strength_level_th6≤v_var＜strength_level_th7, select level7 curve;

When v_var≥strength_level_th7, select level8 curve;

Among them, strength_level_th0~7 represent the threshold values of the preset value range, which can be set according to the needs. The slopes of the corresponding curves level1~8 decrease successively (the absolute values of the slopes increase successively), among which the slope of the level0 curve is 0, that is, no attenuation is performed, and the corresponding β′(x,y) values queried in the curve for each original pixel point (x,y) are all 1.

After the corresponding attenuation curve is selected, the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapped coordinate point as the horizontal coordinate is queried from the selected attenuation curve as the vertical edge correction coefficient of each original pixel point.

In other embodiments, there are multiple alternative solutions to the above-mentioned solution of obtaining the corresponding vertical edge correction coefficient through curve query. For example, the above-mentioned curve can be converted into a corresponding table, and the vertical edge correction coefficient of each original pixel point can be queried and determined by table lookup based on the vertical edge gradient value, vertical edge intensity value and Gaussian distance of each original pixel point relative to the mapping coordinate point of the mapping coordinate point, and the like.

S163, determining an edge enhancement attenuation correction coefficient for each original pixel point according to the horizontal edge correction coefficient and the vertical edge correction coefficient for each original pixel point.

As an example, the edge enhancement attenuation correction coefficient coe _ij of each original pixel may be determined according to the horizontal edge correction coefficient α′(x, y) and the vertical edge correction coefficient β′(x, y) of each original pixel, as shown in the following formula:

coe _ij =α′(x,y)×β′(x,y)

Among them, i is the row subscript of the original pixel point in the neighborhood, j is the column subscript of the original pixel point in the neighborhood, coe _ij is the edge enhancement attenuation correction coefficient of the original pixel point (i, j), and (x, y) is the coordinate value of the original pixel point (i, j) in the neighborhood. Taking the neighborhood including 4×4 original pixels as an example, i can take values 0 to 3, and j can take values 0 to 3.

S170, weighting the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain a convolution filter weight;

As an example, taking the neighborhood including 4×4 original pixels as an example, the weight of each original pixel is weighted by using the edge enhancement attenuation correction coefficient, and the convolution filter weight is obtained as shown in the following formula:

coe₀₀～coe₃₃代表邻域内各原始像素点的边缘增强衰减矫正系数，

Wherein, the subscript ij represents the original pixel corresponding to the i-th row and j-th column in the neighborhood, _w00 ~ _w33 represent the weights of the original pixels in the neighborhood after attenuation enhancement, _wsad00 ~ _wsad33 represent the weights of the original pixels in the neighborhood before attenuation enhancement obtained by calculating the Gaussian distance of the original pixels in the neighborhood,

coe ₀₀ ~ coe ₃₃ represent the edge enhancement attenuation correction coefficients of each original pixel in the neighborhood.

S180, determining the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

As an example, taking the neighborhood including 4×4 original pixels as an example, the pixel value of the pixel to be interpolated in the target image is determined according to the product of the convolution filter weight and the pixel value of each original pixel in the neighborhood as shown in the following formula:

Where pixel _i,j represents the pixel value of the pixel to be interpolated in the i-th row and j-th column in the neighborhood, and p ₀₀ to p ₃₃ represent the pixel values of each original pixel in the 4×4 neighborhood of the pixel to be interpolated.

As shown in Figure 7, the convolution filter window scaling process is schematically shown, where h_step and v_step are the step offsets of the pixels to be interpolated in the horizontal and vertical directions respectively. The pixel to be interpolated is used as the center of the convolution window to search for the original pixels in the 4x4 neighborhood around the mapping coordinates in the original image, and the pixel values of the pixels to be interpolated are obtained after weighted summation and normalization. The above is a method for obtaining a pixel to be interpolated by one convolution. The pixel values of all the pixels to be interpolated in the target image are obtained by the above method, and the process of scaling the original image to obtain the target image is completed.

The above-mentioned image scaling method determines the edge gradient direction information and edge strength information of the mapping coordinate point according to the pixel values of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image corresponding to the pixel point to be interpolated, enhances or attenuates the weight of each original pixel point in the neighborhood to obtain the weight of each original pixel point after enhancement or attenuation, and solves the pixel value of the pixel point to be interpolated according to the weight of each original pixel point after enhancement or attenuation and the pixel value of each original pixel point. It can maintain the correlation between the pixels before and after in the interpolation process, and can perform interpolation processing on the edges in any direction to maintain the edge information of the original image, so that the image remains clear after being enlarged or reduced, avoiding the occurrence of jagged distortion and other phenomena, retaining the integrity and correlation of the image edge information, making the contour and texture of the scaled image clearer, and improving the quality of the target image obtained after scaling.

It should be understood that, although the various steps in the flowcharts of Figures 1-3 are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in Figures 1-3 may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

In one embodiment, as shown in FIG8 , an image scaling device 800 is provided, comprising: a to-be-interpolated pixel acquisition module 810, a neighborhood pixel acquisition module 820, an edge information determination module 830, an original weight determination module 840, an enhanced attenuation correction coefficient determination module 850, and a to-be-interpolated pixel pixel value determination module 860, wherein:

The pixel point acquisition module 810 is used to acquire the pixel points to be interpolated in the target image;

The neighborhood pixel acquisition module 820 is used to determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image;

The edge information determination module 830 is used to determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

The original weight determination module 840 is used to determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapped coordinate point and each original pixel point in the neighborhood;

The enhancement attenuation correction coefficient determination module 850 is used to determine the edge enhancement attenuation correction coefficient of each original pixel in the neighborhood according to the edge gradient direction information, the edge intensity information and the Gaussian distance;

The pixel value determination module 860 of the pixel point to be interpolated is used to weight the weights of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood, the pixel value of the pixel point to be interpolated in the target image is determined.

In one embodiment, the edge information determination module 830 includes:

A horizontal edge information determination module is used to determine the horizontal edge gradient value and the corresponding horizontal edge intensity value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

The vertical edge information determination module is used to determine the vertical edge gradient value and the corresponding vertical edge intensity value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood.

In one embodiment, the horizontal edge information determination module is used to: calculate the mean of the pixel values of each original pixel point in the neighborhood; obtain the original pixel points of a preset row among the original pixel points in the neighborhood; determine the horizontal edge gradient value of the mapping coordinate point according to the difference in the horizontal direction of the original pixel points of the preset row; determine the horizontal edge intensity value of the mapping coordinate point according to the difference between the original pixel points of the preset row and the mean.

In one embodiment, the vertical edge information determination module is used to: calculate the mean of the pixel values of each original pixel point in the neighborhood; obtain the original pixel points of a preset column among the original pixel points in the neighborhood; determine the vertical edge gradient value of the mapping coordinate point based on the difference in the vertical direction of the original pixel points of the preset column; determine the vertical edge intensity value of the mapping coordinate point based on the difference between the original pixel points of the preset column and the mean.

In one embodiment, the enhanced attenuation correction coefficient determination module 850 includes:

A horizontal edge correction coefficient determination module is used to determine the horizontal edge correction coefficient of each original pixel point according to the horizontal edge gradient value, the horizontal edge intensity value and the Gaussian distance of each original pixel point relative to the mapping coordinate point;

A vertical edge correction coefficient determination module is used to determine the vertical edge correction coefficient of each original pixel point according to the vertical edge gradient value, the vertical edge intensity value and the Gaussian distance of each original pixel point relative to the mapping coordinate point;

The edge enhancement attenuation correction coefficient determination module is used to determine the edge enhancement attenuation correction coefficient of each original pixel point according to the horizontal edge correction coefficient and the vertical edge correction coefficient of each original pixel point.

In one embodiment, the horizontal edge correction coefficient determination module is used to: determine the positivity of the horizontal edge enhancement attenuation exponent according to the horizontal edge gradient value of the mapping coordinate point; determine the size of the horizontal edge enhancement attenuation base according to the horizontal edge intensity value of the mapping coordinate point; for each original pixel point, use the horizontal edge enhancement attenuation base as the base and the product of the horizontal edge enhancement attenuation exponent and the Gaussian distance from the original pixel point to the mapping coordinate point as the exponent to perform a power operation to obtain the horizontal edge correction coefficient of the original pixel point.

In one embodiment, the horizontal edge correction coefficient determination module is used to: determine a corresponding curve set according to the positive or negative horizontal edge gradient value of the mapping coordinate point; the curve set includes an enhancement curve set with a curve slope greater than zero or an attenuation curve set with a curve slope less than zero; select a corresponding curve from the determined curve set according to the horizontal edge intensity value, and the horizontal edge intensity value is positively proportional to the absolute value of the slope of the selected corresponding curve; query the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate from the selected curve as the horizontal edge correction coefficient of each original pixel point.

In one embodiment, the vertical edge correction coefficient determination module is used to: determine the positive or negative value of the vertical edge enhancement attenuation exponent according to the vertical edge gradient value of the mapping coordinate point; determine the size of the vertical edge enhancement attenuation base according to the vertical edge strength value of the mapping coordinate point; for each original pixel point, use the vertical edge enhancement attenuation base as the base, and use the product of the vertical edge enhancement attenuation exponent and the Gaussian distance from the original pixel point to the mapping coordinate point as the exponent to perform a power operation to obtain the vertical edge correction coefficient of the original pixel point.

In one embodiment, the vertical edge correction coefficient determination module is used to: determine a corresponding curve set according to the positive or negative vertical edge gradient value of the mapping coordinate point; the curve set includes an enhancement curve set with a curve slope greater than zero or an attenuation curve set with a curve slope less than zero; select a corresponding curve from the determined curve set according to the vertical edge strength value, and the vertical edge strength value is positively proportional to the absolute value of the slope of the selected corresponding curve; query the vertical coordinate corresponding to the Gaussian distance between each original pixel point and the mapping coordinate point as the horizontal coordinate from the selected curve as the vertical edge correction coefficient of each original pixel point.

In one embodiment, the original weight determination module is used to determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood, as shown in the following formula:

Among them, (x, y) is the coordinate value of the original pixel in the neighborhood, r _sd (x, y) is the Gaussian distance between the original pixel and the mapped coordinate point in the neighborhood, w _{(x, y)} is the weight of the original pixel (x, y), n _x is the normalization coefficient, and σ is the standard deviation of the corresponding Gaussian distribution.

In one embodiment, the enhancement attenuation correction coefficient determination module 850 is used to determine the edge enhancement attenuation correction coefficient of each original pixel in the neighborhood according to the edge gradient direction information, the edge intensity information and the Gaussian distance, as shown in the following formula:

In the above formula, coe _ij is the edge enhancement attenuation correction coefficient of the original pixel point in the i-th row and j-th column in the neighborhood, (x, y) is the coordinate value of the original pixel point in the i-th row and j-th column in the neighborhood, r _sd (x, y) is the Gaussian distance of the original pixel point in the i-th row and j-th column in the neighborhood relative to the mapped coordinate point, the positive and negative value of h_signed is determined according to the positive and negative value of the horizontal edge gradient, the positive and negative value of v_signed is determined according to the positive and negative value of the vertical edge gradient, the value of α is positively correlated with the horizontal edge strength value, and the value of β is positively correlated with the vertical edge strength value.

For the specific definition of the image scaling device, please refer to the definition of the image scaling method above, which will not be repeated here. Each module in the above image scaling device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG9. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, an image scaling method is implemented. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

Those skilled in the art will understand that the structure shown in FIG. 9 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

Obtain the pixel point to be interpolated in the target image; determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image; determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood; determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood; determine the edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance; weight the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; determine the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

In other embodiments, when the processor executes the computer program, the steps of the image scaling method in any one of the above embodiments are also implemented.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

Obtain the pixel point to be interpolated in the target image; determine the mapping coordinate point of the pixel point to be interpolated in the original image, and obtain the pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image; determine the edge gradient direction information and edge intensity information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood; determine the weight of each original pixel point in the neighborhood according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood; determine the edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, edge intensity information and Gaussian distance; weight the weight of each original pixel point using the edge enhancement attenuation correction coefficient to obtain the convolution filter weight; determine the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filter weight and the pixel value of each original pixel point in the neighborhood.

In other embodiments, when the computer program is executed by a processor, the steps of the image scaling method of any of the above embodiments are also implemented.

Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims (15)

1. A method of image scaling, the method comprising:

acquiring a pixel point to be interpolated in a target image;

determining a mapping coordinate point of the pixel point to be interpolated in the original image, and acquiring a pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image;

determining edge gradient direction information and edge strength information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood, wherein the edge gradient direction information comprises a horizontal edge gradient value and a vertical edge gradient value, the edge strength information comprises a horizontal edge strength value and a vertical edge strength value,

selecting original pixel points in a neighborhood of 4x4 around the mapping coordinate point, wherein the mean value of pixel values of all the original pixel points in the neighborhood is calculated as follows:

avg is the average value of the pixel values of all original pixel points in the neighborhood, i is the subscript of the row of the original pixel points in the neighborhood, i takes the value of 0-3,j as the subscript of the row of the original pixel points in the neighborhood, and j takes the value of 0-3,p _i,j The pixel value of the original pixel point (i, j) in the neighborhood,

the horizontal edge strength value is calculated as follows:

wherein h _ var is the horizontal edge strength value, abs is the absolute value,

the vertical edge strength value is calculated as follows:

in the formula, v _ var is a vertical edge strength value, and abs is an absolute value;

determining the weight of each original pixel point in the neighborhood by adopting the following formula according to the Gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood:

wherein, (x, y) is the coordinate value of the original pixel point in the neighborhood, r _sd (x, y) is the Gaussian distance between the original pixel point with the coordinate value (x, y) in the neighborhood and the mapping coordinate point, w _(x,y) Is the weight of the original pixel point with coordinate value of (x, y), n _x Is a normalized coefficient, and σ is a standard deviation of the corresponding gaussian distribution;

determining an edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, the edge strength information and the Gaussian distance, wherein the edge enhancement attenuation correction coefficient is calculated by adopting the following formula:

of these, coe _ij The method comprises the steps that an edge enhancement attenuation correction coefficient of the ith row and the jth column of original pixel points in a neighborhood is calculated, h _ signed is a horizontal edge enhancement attenuation index, the positive and negative of h _ signed is determined according to the positive and negative of a horizontal edge gradient value, v _ signed is a vertical edge enhancement attenuation index, the positive and negative of v _ signed is determined according to the positive and negative of a vertical edge gradient value, alpha is a horizontal edge enhancement attenuation base number, the value size of alpha is in positive correlation with a horizontal edge strength value, beta is a vertical edge enhancement attenuation base number, and the value size of beta is in positive correlation with a vertical edge strength value;

weighting the weight of each original pixel point by using the edge enhancement attenuation correction coefficient to obtain a convolution filtering weight; and determining the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filtering weight and the pixel value of each original pixel point in the neighborhood.

2. The method according to claim 1, wherein the determining edge gradient direction information and edge strength information of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood comprises:

determining a horizontal edge gradient value and a corresponding horizontal edge intensity value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

and determining the vertical edge gradient value and the corresponding vertical edge strength value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood.

3. The method of claim 2, wherein determining the horizontal edge gradient value and the corresponding horizontal edge strength value for the mapped coordinate point according to the pixel values of the original pixels in the neighborhood comprises:

calculating the mean value of the pixel values of all the original pixel points in the neighborhood;

acquiring original pixel points of a preset row in the original pixel points in the neighborhood;

determining a horizontal edge gradient value of the mapping coordinate point according to the difference value of the original pixel points of the preset row in the horizontal direction;

and determining the horizontal edge intensity value of the mapping coordinate point according to the difference value between the original pixel point of the preset row and the mean value.

4. The method as claimed in claim 2, wherein determining the vertical edge gradient value and the corresponding vertical edge strength value of the mapping coordinate point according to the pixel values of the original pixels in the neighborhood includes:

calculating the mean value of the pixel values of all the original pixel points in the neighborhood;

acquiring original pixel points of a preset column in the original pixel points in the neighborhood;

determining a vertical edge gradient value of the mapping coordinate point according to the difference value of the original pixel points of the preset column in the vertical direction;

and determining the vertical edge strength value of the mapping coordinate point according to the difference value between the original pixel point of the preset column and the average value.

5. The method according to any one of claims 1 to 4, wherein the determining an edge enhancement attenuation correction coefficient of each original pixel point in the neighborhood according to the edge gradient direction information, the edge strength information and the Gaussian distance comprises:

determining a horizontal edge correction coefficient of each original pixel point according to the horizontal edge gradient value and the horizontal edge intensity value of the mapping coordinate point and the Gaussian distance of each original pixel point relative to the mapping coordinate point;

determining a vertical edge correction coefficient of each original pixel point according to the vertical edge gradient value and the vertical edge strength value of the mapping coordinate point and the Gaussian distance of each original pixel point relative to the mapping coordinate point;

and determining the edge enhancement attenuation correction coefficient of each original pixel point according to the horizontal edge correction coefficient and the vertical edge correction coefficient of each original pixel point.

6. The method of claim 5, wherein determining the horizontal edge rectification coefficients for each raw pixel point based on the horizontal edge gradient values, the horizontal edge intensity values, and the Gaussian distance of each raw pixel point from the mapped coordinate point comprises:

determining the positive and negative of a horizontal edge enhancement attenuation index according to the horizontal edge gradient value of the mapping coordinate point;

determining the magnitude of the horizontal edge enhancement attenuation base number according to the horizontal edge strength value of the mapping coordinate point;

aiming at each original pixel point, taking the horizontal edge enhancement attenuation base number as the base number, taking the product of the horizontal edge enhancement attenuation index and the Gaussian distance from the original pixel point to the mapping coordinate point as the index, and performing power operation to obtain the horizontal edge correction coefficient of the original pixel point.

7. The method of claim 5, wherein determining the horizontal edge rectification coefficient for each raw pixel point according to the horizontal edge gradient value, the horizontal edge intensity value and the Gaussian distance of each raw pixel point from the mapped coordinate point comprises:

determining a corresponding curve set according to the positive and negative of the horizontal edge gradient value of the mapping coordinate point; the curve set comprises an enhancement curve set with a curve slope larger than zero or an attenuation curve set with a curve slope smaller than zero;

selecting a corresponding curve from the determined curve set according to the horizontal edge intensity value, wherein the horizontal edge intensity value is in direct proportion to the absolute value of the slope of the selected corresponding curve;

and inquiring a vertical coordinate corresponding to the abscissa by taking the Gaussian distance between each original pixel point and the mapping coordinate point as the abscissa from the selected curve, and taking the vertical coordinate as the horizontal edge correction coefficient of each original pixel point.

8. The method of claim 5, wherein determining the vertical edge rectification coefficient for each original pixel point according to the vertical edge gradient value, the vertical edge strength value and the Gaussian distance of each original pixel point to the mapping coordinate point comprises:

determining the positive and negative of a vertical edge enhanced attenuation index according to the vertical edge gradient value of the mapping coordinate point;

determining the enhancement attenuation base number of the vertical edge according to the vertical edge strength value of the mapping coordinate point;

and aiming at each original pixel point, performing power operation by taking the vertical edge enhanced attenuation base number as the base number and taking the product of the vertical edge enhanced attenuation index and the Gaussian distance from the original pixel point to the mapping coordinate point as the index to obtain the vertical edge correction coefficient of the original pixel point.

9. The method of claim 5, wherein determining the vertical edge rectification coefficient for each original pixel point according to the vertical edge gradient value and the vertical edge intensity value of the mapped coordinate point and the Gaussian distance of each original pixel point relative to the mapped coordinate point comprises:

determining a corresponding curve set according to the positive and negative of the vertical edge gradient value of the mapping coordinate point; the curve set comprises an enhancement curve set with a curve slope larger than zero or an attenuation curve set with a curve slope smaller than zero;

selecting a corresponding curve from the determined curve set according to the vertical edge strength value, wherein the vertical edge strength value is in direct proportion to the absolute value of the slope of the selected corresponding curve;

and inquiring a vertical coordinate corresponding to an abscissa by taking the Gaussian distance between each original pixel point and the mapping coordinate point as the abscissa from the selected curve, and taking the vertical coordinate as a vertical edge correction coefficient of each original pixel point.

10. An image scaling apparatus, characterized in that the apparatus comprises:

the to-be-interpolated pixel point acquisition module is used for acquiring a to-be-interpolated pixel point in the target image;

the neighborhood pixel acquisition module is used for determining a mapping coordinate point of the pixel point to be interpolated in the original image and acquiring a pixel value of each original pixel point in a preset neighborhood around the mapping coordinate point in the original image;

an edge information determining module, configured to determine, according to a pixel value of each original pixel point in the neighborhood, edge gradient direction information and edge strength information of the mapping coordinate point, where the edge gradient direction information includes a horizontal edge gradient value and a vertical edge gradient value, and the edge strength information includes a horizontal edge strength value and a vertical edge strength value; selecting a 4x4 neighborhood around the mapped coordinate pointThe average value of the pixel values of all the original pixel points in the neighborhood is calculated as follows:

avg is the average value of the pixel values of all original pixel points in the neighborhood, i is the subscript of the row of the original pixel points in the neighborhood, i takes the value of 0-3,j as the subscript of the row of the original pixel points in the neighborhood, and j takes the value of 0-3,p _i,j The pixel value of the original pixel point (i, j) in the neighborhood; the horizontal edge strength value is calculated as follows:

in the formula, h _ var is a horizontal edge strength value, and abs is an absolute value; the vertical edge strength value is calculated as follows:

in the formula, v _ var is a vertical edge strength value, and abs is an absolute value;

an original weight determining module, configured to determine, according to a gaussian distance between the mapping coordinate point and each original pixel point in the neighborhood, a weight of each original pixel point in the neighborhood by using the following formula:

wherein, (x, y) is the coordinate value of the original pixel point in the neighborhood, r _sd (x, y) is the Gaussian distance between the original pixel point with the coordinate value (x, y) in the neighborhood and the mapping coordinate point, w _(x,y) Is the weight of the original pixel point with coordinate value of (x, y), n _x Is a normalized coefficient, and σ is a standard deviation of the corresponding gaussian distribution;

an enhanced attenuation correction coefficient determining module, configured to determine, according to the edge gradient direction information, the edge strength information, and the gaussian distance, an edge enhanced attenuation correction coefficient of each original pixel point in the neighborhood, where the edge enhanced attenuation correction coefficient is calculated by using the following formula:

wherein coe _ij The method comprises the steps that an edge enhancement attenuation correction coefficient of the ith row and the jth column of original pixel points in a neighborhood is calculated, h _ signed is a horizontal edge enhancement attenuation index, the positive and negative of h _ signed is determined according to the positive and negative of a horizontal edge gradient value, v _ signed is a vertical edge enhancement attenuation index, the positive and negative of v _ signed is determined according to the positive and negative of a vertical edge gradient value, alpha is a horizontal edge enhancement attenuation base number, the value size of alpha is in positive correlation with a horizontal edge strength value, beta is a vertical edge enhancement attenuation base number, and the value size of beta is in positive correlation with a vertical edge strength value;

the pixel value determining module of the pixel point to be interpolated is used for weighting the weight of each original pixel point by utilizing the edge enhancement attenuation correction coefficient to obtain the convolution filtering weight; and determining the pixel value of the pixel point to be interpolated in the target image according to the product of the convolution filtering weight and the pixel value of each original pixel point in the neighborhood.

11. The apparatus of claim 10, wherein the edge information determining module comprises:

the horizontal edge information determining module is used for determining a horizontal edge gradient value and a corresponding horizontal edge strength value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood;

and the vertical edge information determining module is used for determining the vertical edge gradient value and the corresponding vertical edge strength value of the mapping coordinate point according to the pixel value of each original pixel point in the neighborhood.

12. The apparatus according to claim 11, wherein the horizontal edge information determining module is configured to calculate a mean value of pixel values of original pixels in the neighborhood; acquiring original pixel points of a preset row in the original pixel points in the neighborhood; determining a horizontal edge gradient value of the mapping coordinate point according to the difference value of the original pixel points of the preset row in the horizontal direction; and determining the horizontal edge intensity value of the mapping coordinate point according to the difference value between the original pixel point of the preset row and the mean value.

13. The apparatus of claim 11, wherein the vertical edge information determining module is configured to calculate a mean value of pixel values of original pixels in the neighborhood; acquiring original pixel points of a preset column in the original pixel points in the neighborhood; determining a vertical edge gradient value of the mapping coordinate point according to a difference value of the original pixel points of the preset column in the vertical direction; and determining the vertical edge strength value of the mapping coordinate point according to the difference value between the original pixel point of the preset column and the average value.

14. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor realizes the steps of the image scaling method according to any of claims 1 to 9 when executing the computer program.

15. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the image scaling method according to any one of claims 1 to 9.

Images