TWI864390B - Image processing apparatus for video adjustment and video enhancement method - Google Patents

Abstract

An image processing apparatus for video adjustment and a video enhancement method are provided. In the method, an image frame of a video is divided into one or more facial blocks. The facial blocks are corresponding to a face portion captured in the image frame. The brightness of one or more pixels in each facial block is adjusted according to a converted curve. The initial brightness of one pixel has a corresponding adjusted brightness in the converted curve. The converted curve is related to a polynomial, which is formed according to the initial brightness and the adjusted brightness. The adjusted level of a pixel is adjusted according to the primary color ratio of the pixel in the facial block. The primary color ratio is the ratio of the adjusted level of one primary color in the facial block. The adjusted level is the level of the primary color to form the adjusted brightness. Accordingly, a lighting effect is provided.

Description

Image processing device for adjusting video and video enhancement method

The present invention relates to an image processing technology, and in particular to an image processing device and a video enhancement method for adjusting video.

Remote video conferencing allows people in different locations or spaces to have conversations, and conference-related equipment, protocols and/or applications have also developed quite maturely. In addition, live video broadcasting has also been a very popular industry in recent years, and live video broadcasting allows viewers, fans or customers to watch the live broadcaster's conversation content or shared screens in real time.

Inevitably, the user or live streamer may be in an environment with insufficient light, which reduces the visual experience of the viewer on the other end. Some users or recording devices will set up light-emitting elements to provide additional light to the face, other body parts or the environment. However, additional light-emitting elements will increase hardware costs and even cause problems such as uneven brightness distribution and color temperature deviation.

In view of this, the present invention provides an image processing device and a video enhancement method for adjusting the video, which directly adjusts the brightness of the image in the video to achieve a lighting effect.

The video enhancement method of the embodiment of the present invention includes (but is not limited to) the following steps: segmenting the video image frame into one or more face blocks. The face block corresponds to the face captured by the image frame. Adjusting the brightness value of one or more pixels in each face block according to the conversion curve. The initial brightness value of a pixel corresponds to an adjusted brightness value in the conversion curve, and the conversion curve is related to a polynomial formed according to the initial brightness value and the adjusted brightness value. Adjusting the adjusted color value of the pixel according to the primary color ratio of the pixel in the face block. The primary color ratio is the proportion of the adjusted color value of a primary color in a face block. The adjusted color value is the color value of the primary color that forms the adjusted brightness value.

The image processing device of the embodiment of the present invention includes (but is not limited to) a memory and a processor. The memory is used to store program code. The processor is coupled to the memory. The processor is configured to load and execute the program code to divide the image frame of the video into one or more face blocks, adjust the brightness value of one or more pixels in each face block according to the transformation curve, and adjust the adjusted color level value of the pixel according to the primary color ratio of the pixel in the face block. The face block corresponds to the face captured by the image frame. The initial brightness value of a pixel corresponds to the adjusted brightness value in the transformation curve, and the transformation curve is related to a polynomial formed according to the initial brightness value and the adjusted brightness value. The primary color ratio is the proportion of the color level value of a primary color in a face block. The adjusted chromaticity value is the chromaticity value of the primary color that forms the adjusted brightness value.

Based on the above, according to the image processing device and video enhancement method for adjusting video according to the embodiment of the present invention, the brightness of one or more face blocks in the image frame is smoothly adjusted using an adjustment curve, and the color scale value is fine-tuned according to the primary color ratio. In this way, a natural and evenly distributed brightness can be provided, thereby improving the viewing experience. In addition, the embodiment of the present invention can avoid the need for additional light-emitting elements.

In order to make the above features and advantages of the present invention more clearly understood, the following is a detailed description of the embodiments with the accompanying drawings.

100: Electronic devices

100: Image processing device

110: Storage

111: Initialize module

113: Brightness adjustment module

115: Ratio adjustment module

117: Differential adjustment module

130: Processor

S210~S250: Steps

FP: Facial feature points

IF1, IF2: Image frame

RB1~RB4: Regional block

PB1~PB3: Personalized area

L _brt : adjusted color gradation value

_Lin : Initial color level value

TL, LL, BL, P: Position

SF:S function

FIG1 is a block diagram of components of an image processing device according to an embodiment of the present invention.

Figure 2 is a flow chart of a video enhancement method according to an embodiment of the present invention.

Figure 3 is a schematic diagram of facial feature points according to an embodiment of the present invention.

FIG4 is a schematic diagram of a facial block-regional block according to an embodiment of the present invention.

Figure 5 is a schematic diagram of a facial block-personalized block according to an embodiment of the present invention.

Figure 6 is a schematic diagram of an adjustment curve according to an embodiment of the present invention.

Figure 7 is a schematic diagram of the S function (Sigmoid) according to an embodiment of the present invention.

FIG1 is a block diagram of components of an image processing device 100 according to an embodiment of the present invention. Referring to FIG1 , the image processing device 100 includes (but is not limited to) a memory 110 and a processor 130. The image processing device 100 may be a desktop computer, a laptop computer, a smart phone, a tablet computer, a server or other computing devices.

The memory 110 can be any type of fixed or removable random access memory (RAM), read only memory (ROM), flash memory, traditional hard disk drive (HDD), solid-state drive (SSD) or similar components. In one embodiment, the memory 110 is used to record program code, software modules (e.g., initialization module 111, brightness adjustment module 113, ratio adjustment module 115 and difference adjustment module 117), configuration, data (e.g., images, values, feature points, curves, blocks, etc.) or files, and its embodiments will be described in detail later.

The processor 130 is coupled to the memory 110, and the processor 130 may be a central processing unit (CPU), a graphic processing unit (GPU), or other programmable general-purpose or special-purpose microprocessor, digital signal processor (DSP), programmable controller, field programmable gate array (FPGA), application-specific integrated circuit (ASIC), neural network accelerator or other similar components or a combination of the above components. In one embodiment, the processor 130 is used to execute all or part of the operations of the image processing device 100. The processor 130 can load and execute various program codes, software modules, files and data stored in the memory 110.

In the following, the method described in the embodiment of the present invention will be described with various devices, components and modules in the image processing device 100. The various processes of the method can be adjusted according to the implementation situation, but are not limited to this.

FIG2 is a flow chart of a video enhancement method according to an embodiment of the present invention. Referring to FIG2, the processor 130 divides the video image frame into one or more face blocks through the initialization module 111 (step S210). Specifically, the video is a multimedia stream obtained by recording a person, a specific target object or a scene. A collection of one or more image frames can be referred to as a video. In one embodiment, the processor 130 can obtain the video through a built-in or external image capture device (e.g., a camera, a camera or a monitor). In another embodiment, the processor 130 can download the video from a server, a computer or a storage medium.

Generally speaking, the subject of a video conference or live broadcast is a human face. That is, the image capture device faces the human face. The face block corresponds to the face captured by the image frame. In other words, the initialization module 111 divides the face into one or more face regions. These face regions are used for subsequent zoning processing of brightness or color level adjustment.

There are many ways to segment the image frame. In one embodiment, the initialization module 111 can segment the image frame according to one or more facial feature points in the image frame. One or more facial blocks include one or more facial feature points. Facial feature points may be organs, contours, edges or specific locations on the face. For example, the corners of the eyes, the corners of the mouth, the tip of the nose, etc. In other words, the segmentation of the facial block is related to the location of the facial feature points.

For the recognition of facial feature points, in one embodiment, the initialization module 111 can use a recognition module trained based on a machine learning algorithm (e.g., Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Multi-Layer Perceptron (MLP) or Support Vector Machine (SVM)) to recognize these facial feature points. The machine learning algorithm can analyze the training samples to obtain rules from them, thereby predicting unknown data through the rules. The recognition module is a machine learning model constructed after learning, and is used to recognize or classify facial feature points in the image frame.

For example, FIG3 is a schematic diagram of facial feature points FP according to an embodiment of the present invention. Referring to FIG3, facial landmark points FP of 68 can be marked by a facial landmark model based on machine learning. Several facial landmark points FP can be connected to form contours of areas such as eyes, eyebrows, nose, mouth, chin or forehead.

In another embodiment, the initialization module 111 may use techniques such as Harr features, Speeded Up Robust Features (SURF), scale-invariant feature transform (SIFT), Adaboost or other image recognition and/or feature matching techniques to identify facial key points.

In one embodiment, the face block includes a plurality of face blocks, and a boundary of one of the face blocks is adjacent to a boundary of another of the face blocks. For example, the face block includes a region block. FIG. 4 is a schematic diagram of a face block-region block RB1-RB4 according to an embodiment of the present invention. Referring to FIG. 4, the initialization module 111 divides the image frame IF1 into four region blocks RB1-RB4, and divides the face F into upper left, upper right, lower left and lower right blocks accordingly. Among them, the right boundary of the regional block RB1 is adjacent to the left boundary of the regional block RB2, and the lower boundary of the regional block RB1 is adjacent to the upper boundary of the regional block RB3, and so on. The several facial feature points FP marked on the outer contour of the face F will be assigned to the regional blocks corresponding to their positions.

It should be noted that the segmentation size, number and position shown in Figure 4 are only for illustrative purposes and can be changed according to design requirements. In addition, the sizes of different facial blocks may also be different.

In one embodiment, one or more facial feature points in one or more facial blocks correspond to organs in the face (e.g., eyes, nose, or mouth). In other words, these facial regions are targeted at specific facial organs. For example, the facial block includes a personalized block. FIG5 is a schematic diagram of a facial block-personalized block PB1~PB3 according to an embodiment of the present invention. Referring to FIG5, the initialization module 111 divides the image frame IF2 into three personalized blocks PB1~PB3. Among them, the personalized block PB1 frames the facial feature points FP of the eyes. The personalized block PB2 frames the facial feature points FP of the nose. The personalized block PB3 frames the facial feature points FP of the mouth.

It should be noted that the size, number and location of the partitions shown in Figure 5 are only for illustrative purposes and can be changed according to design requirements. In addition, any two personalized blocks may partially overlap, be adjacent or distant.

In some embodiments, the initialization module 111 can segment the image frame into two regions and personalized regions. In addition, in other embodiments, the segmentation of the image frame is not limited to the position of the reference facial feature points. For example, the initialization module 111 segments the image frame according to the geometric center and contour of the face.

Referring to FIG. 2 , the processor 130 adjusts the brightness value of one or more pixels in each face block according to the conversion curve through the brightness adjustment module 113 (step S230). Specifically, the conversion curve is related to a polynomial formed according to the initial brightness value and the adjusted brightness value. The initial brightness value represents the original brightness value of a pixel in the image frame, and the adjusted brightness value represents that the brightness value of the pixel has been adjusted (may be the same or different from the original brightness value). The domain of the polynomial is, for example, the lowest value to the highest value of the initial brightness value, and its range is, for example, the lowest value to the highest value of the adjusted brightness value. The initial brightness value of the pixel corresponds to the adjusted brightness value in the conversion curve.

T_bj為第j臉部區塊的亮度代表值，R_i為第j臉部區塊中的第i像素的紅色色階值，G_i為第i像素的綠色色階值，B_i為第i像素的藍色色階值，且P為第j臉部區塊中的像素總數。初始化模組111可定義目標亮度值。這目標亮度值可能是所有臉部區塊的亮度值的統計指標或其他合適於觀賞的亮度值。 In one embodiment, the polynomial is a quadratic curve equation. The initialization module 111 may obtain a brightness representative value of one or more pixels in the face block. The brightness representative value is related to a statistical index (e.g., mean, median, or mode) of the brightness values of these pixels. For example, for a certain face block, the initialization module 111 calculates the arithmetic mean of the color values of the three primary colors (i.e., red, green, and blue) of all pixels:

T _bj is the brightness representative value of the j-th facial block, _Ri is the red color value of the i-th pixel in the j-th facial block, _Gi is the green color value of the i-th pixel, _Bi is the blue color value of the i-th pixel, and P is the total number of pixels in the j-th facial block. The initialization module 111 can define a target brightness value. This target brightness value may be a statistical indicator of the brightness values of all facial blocks or other brightness values suitable for viewing.

It is worth noting that the brightness representative value and the target brightness value can pass through the quadratic curve equation. FIG. 6 is a schematic diagram of an adjustment curve according to an embodiment of the present invention. Referring to FIG. 6, in terms of a two-dimensional rectangular coordinate system, the adjustment curve formed by the quadratic curve equation can pass through the position TL of the coordinate (brightness representative value, target brightness value). In addition, the adjustment curve further passes through the position LL of the coordinate (brightness minimum value, brightness minimum value) and the position BL of the coordinate (brightness maximum value, brightness maximum value). For example, the brightness minimum value is zero, and the brightness maximum value is 255. The initialization module 111 can input these coordinates into the quadratic equation whose coefficients have not yet been determined, and determine its coefficients accordingly. For example, the mathematical expression of the quadratic curve equation is: L _brt =((T _bj -T _aj )/(T _aj ² -k * T _aj )) * _Lin ² +((1-k * (T _bj -T _aj )/(T _aj -k * T _aj )) * _Lin ...(2)L _brt is the adjusted color level value of a primary color of a pixel in the j-th facial block, Lin _is the initial color level value of the primary color of the pixel, T _bj is the target brightness value of the j-th facial block, T _aj is the representative brightness value of the j-th facial block, and k is the highest color level value (e.g., 255).

The brightness adjustment module 113 can obtain the adjusted brightness values of the pixels in these face blocks according to the color scale values of one or more pixels in each face block and the quadratic curve equation. For example, the brightness adjustment module 113 substitutes the color scale value of a primary color of a certain pixel into the aforementioned equation (2) and obtains the adjusted color scale value of the primary color accordingly. The brightness adjustment module 113 determines the adjusted brightness value of the pixel according to the adjusted color scale values of the three primary colors of the pixel. For example, in the conversion between the RGB (red, green, blue) color space and the HSV (hue, saturation, brightness) color space, the brightness (i.e., the brightness value) is the average value of the highest and lowest color scale values of the three primary colors. That is, the adjusted color scale value is the color scale value of the primary color that forms the adjusted brightness value. The adjusted brightness values of other facial blocks and/or pixels can be deduced in the same way.

In another embodiment, the polynomial is not limited to the quadratic equation and can be passed through other specified coordinate points. For example, the brightness adjustment module 113 further adjusts the median or mode of the brightness value of the facial area to another target brightness value.

2, the processor 130 adjusts the adjusted color value of the pixel according to the primary color ratio of the pixel in the brightness adjusted block through the ratio adjustment module 115 (step S250). Specifically, the primary color ratio is the proportion of the adjusted color value of a primary color (e.g., red, green or blue) in a face block. The mathematical expression of the primary color ratio is:

R_ai為第j臉部區塊中的第i像素的紅色的經調整色階值(例如，經調整色階值L_brt)，G_ai為第i像素的綠色的經調整色階值，B_ai為第i像素的藍色的經調整色階值。R_bj為第j臉部區塊的紅色的原色比值，G_bj為綠色的原色比值，B_bj為藍色的原色比值。

_Rai is the adjusted color gradation value of red of the i-th pixel in the j-th face block (e.g., the adjusted color gradation value _Lbrt ), _Gai is the adjusted color gradation value of green of the i-th pixel, and _Bi is the adjusted color gradation value of blue of the i-th pixel. _Rbj is the primary color ratio of red of the j-th face block, _Gbj is the primary color ratio of green, and _Bbj is the primary color ratio of blue.

The ratio adjustment module 115 defines a target ratio. This target ratio may be a statistical indicator of the ratio of all facial blocks or other ratios suitable for viewing. For example, increasing the primary color ratios R _bj and G _bj of yellow skin can make the skin yellower, and increasing the primary color ratio B _bj can make the skin whiter. In some embodiments, the processor 130 can further optimize the organs corresponding to the personalized blocks. For example, for the mouth, the processor 130 can increase the primary color ratio R _bj to make the lips redder; for the eyes, the processor 130 can simultaneously increase the primary color ratios R _bj , G _bj , and B _bj to reduce dark circles.

L_rgb為第j臉部區塊中的某一像素的某一原色的原色比值經調整後的微調色階值，L_brt為經調整色階值，V_bj為第j臉部區塊的相同原色的目標比值，V_aj為第j臉部區塊的相同原色的原色比值，k為色階最大值(例如，255)，C_b為S函數輸入調整係數(假設輸出飽和區約為輸入為+-5，則輸入調整系數為0.2(1/5=0.2)，但不以此為限)，C_a為正規化增益的位移量(用以抵銷S函數的輸出偏移，但可視情況省略)。 The ratio adjustment module 115 adjusts the primary color ratio of each facial block to a target ratio as a target to adjust the pixels in each facial block. In one embodiment, the ratio adjustment module 115 can adjust the primary color ratio to the target ratio according to the characteristics of the S function (Sigmoid). It is worth noting that the characteristic of the S function is that when the value in the definition domain is closer to negative infinity, the corresponding value in the value range tends to zero; when the value in the definition domain is closer to positive infinity, the corresponding value in the value range tends to one. Based on this characteristic, the ratio adjustment module 115 sets the primary color ratio to the definition domain of the S function, and sets the target ratio to the value range of the S function. For example, the mathematical expression of the S function is:

L _rgb is the fine-tuned color gradation value after adjustment of the primary color ratio of a certain primary color of a certain pixel in the j-th facial block, L _brt is the adjusted color gradation value, V _bj is the target ratio of the same primary color of the j-th facial block, V _aj is the primary color ratio of the same primary color of the j-th facial block, k is the maximum color gradation value (for example, 255), C _b is the S function input adjustment coefficient (assuming that the output saturation zone is approximately +-5 when the input is +-5, the input adjustment coefficient is 0.2 (1/5=0.2), but not limited to this), and _Ca is the displacement of the normalized gain (used to offset the output offset of the S function, but can be omitted as needed).

In addition, FIG. 7 is a schematic diagram of an S function SF according to an embodiment of the present invention. Referring to FIG. 7, the input-output relationship of the S function SF is shown in FIG. 7. For example, position P indicates that the input adjusted color scale value is close to the maximum value (for example, the adjusted color scale value is 4.25), and the output fine-tuning color scale value is close to 1 (for example, the fine-tuning color scale value is 0.9859). In this way, the conversion curve can be prevented from causing the output color to be too dark or too light.

It should be noted that the adjustment of the primary color ratio is not limited to the S function characteristics, and other functions that limit the maximum or minimum values can also be applied.

In one embodiment, the colors of the borders between different facial blocks may differ greatly. The processor 130 adjusts the color values of a plurality of adjacent pixels at the borders between two facial blocks through the difference adjustment module 117 to reduce the difference between the color values of those adjacent pixels. The border represents the overlapping or connected portion of the two facial blocks. The adjacent pixels represent one or more pixels within a specific range from the border. Since the aforementioned brightness and ratio adjustments are performed for each facial block individually, the conversion curves and/or primary color ratios used for different facial blocks may be different, and some pixels at the borders of the blocks may have obvious color difference. The difference adjustment module 117 performs blurring/smoothing on these adjacent pixels to reduce the color level difference.

由此可知，相距交界處越近，則距離權重越高；相距交界處越遠， 則距離權重越低。 There are many ways to reduce the difference. In one embodiment, the difference adjustment module 117 can assign distance weights and block weights to those neighboring pixels. The distance weight is inversely proportional to the distance from the junction, and the block weight is related to the importance of the organ in the facial block. For example, Table (1) shows the correspondence between the distance weight and the distance from the junction:

From this we can see that the closer the distance is to the junction, the higher the distance weight is; the farther the distance is to the junction, the lower the distance weight is.

由此可知，嘴巴的重要性較高，但不以此為限。 Table (2) shows the correspondence between block weights and facial blocks:

From this we can see that the mouth is more important, but it is not limited to this.

The difference adjustment module 117 may determine the color values of the neighboring pixels according to the distance weights and block weights of the neighboring pixels. For example, the difference adjustment module 117 calculates a weighted average for a certain pixel and one or more neighboring pixels in a specific direction according to the distance weight and the block weight, and uses the weighted average to determine the color level value after difference adjustment: L _out =(ΣW _dul * W _tul * L _ul +ΣW _dur * W _tur * L _ur +ΣW _ddl * W _tdl * L _dl +ΣW _ddr * W _tdr * L _dr +W _tin * L _in )/(ΣW _dul * W _tul +ΣW _dur * W _tur +ΣW _tdl * W _ddl +ΣW _ddr * W _tdr +W _tin )...(7) L _out is the color level value of a certain primary color of a certain pixel after difference adjustment, L _in is the fine-tuned color value of the pixel (i.e., the color value after ratio adjustment), _Lu1 , _Lu2 , _Ldl , _Ldr are the fine-tuned color values of the neighboring pixels at the upper left, upper right, lower left, and lower right of the pixel, _Wd is the distance weight between the pixel and the boundary (as shown in Table (1)), _Wtin is the block weight of the face block where the pixel is located (as shown in Table (2)), _Wtul , _Wtur , _Wtdl , _Wtdr are the block weights of the neighboring face blocks of the face block where the pixel is located (as shown in Table (2)).

It should be noted that the aforementioned specific directions are not limited to the upper left, upper right, lower left and lower right, and the upper, left, right, lower or other directions may also be applied.

The image frames formed by adjusting the brightness, ratio and/or difference of the face blocks can be output. For example, the image frames are output via network packets or displayed on a monitor. In addition, the video formed by one or more image frames can be played back to improve the brightness of the image frames displayed on the monitor.

In summary, in the image processing device and video enhancement method for adjusting video of the embodiment of the present invention, the brightness, ratio and/or difference are optimized in different regions to achieve the lighting effect. Even if the head moves or the light source moves, the brightness in the image frame can still be evenly distributed and natural. In addition, by using software lighting, the hardware cost of setting up light-emitting elements can be avoided.

Although the present invention has been disclosed as above by the embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of the attached patent application.

S210~S250: Steps

Claims (14)

A video enhancement method includes: dividing an image frame of a video into at least one face block, wherein the at least one face block corresponds to a face captured by the image frame; adjusting the brightness value of at least one pixel in each of the face blocks according to a conversion curve, wherein an initial brightness value of the pixel corresponds to an adjusted brightness value in the conversion curve, the conversion curve is related to a polynomial formed according to the initial brightness value and the adjusted brightness value, and adjusting the brightness value of at least one pixel in each of the face blocks according to the conversion curve. The step of adjusting the brightness value of at least one pixel in the at least one face block comprises: substituting an initial color scale value of each of the three primary colors of one of the at least one pixel in each of the face blocks into the polynomial to obtain adjusted color scale values of the three primary colors; and determining an adjusted brightness value according to the adjusted color scale values of the three primary colors, wherein the adjusted color scale value of each of the primary colors is a color scale value of one of the primary colors forming the adjusted brightness value; and adjusting a primary color ratio of the at least one pixel in the at least one face block to a target ratio to determine A fine-tuned color scale value of the at least one pixel, wherein the primary color ratio of the at least one pixel is the ratio of the adjusted color scale value of the primary color to the sum of the adjusted color scale values of the three primary colors in the face block, and the step of adjusting the primary color ratio of the at least one pixel in the at least one face block to the target ratio includes: adjusting the primary color ratio to the target ratio according to a Sigmoid function characteristic, wherein the adjusted color scale value of the primary color and a first ratio are related to the definition of a Sigmoid function domain, the first ratio is the ratio of the target ratio to the primary color ratio, the fine-tuning color scale value of the primary color is related to the value range of the S function, and the S function characteristics include: when the value in the definition domain of the S function is closer to negative infinity, the corresponding value in the value range of the S function tends to zero; and when the value in the definition domain of the S function is closer to positive infinity, the corresponding value in the value range of the S function tends to one, and when the adjusted color scale value of the primary color is closer to a maximum value, the fine-tuning color scale value of the primary color is closer to one. The video enhancement method as described in claim 1, wherein the polynomial is a quadratic curve equation, and the step of adjusting the brightness value of the at least one pixel in each of the face blocks according to the conversion curve includes: obtaining a brightness representative value of the at least one pixel in the face block, wherein the brightness representative value is related to a statistical indicator of the brightness value of the at least one pixel, and the adjustment curve formed by the quadratic curve equation in a two-dimensional rectangular coordinate system passes through the coordinates of the brightness representative value and a target brightness value in the two-dimensional rectangular coordinate system; and obtaining the adjusted brightness value of the at least one pixel in the at least one face block according to the color value of the at least one pixel in each of the face blocks and the quadratic curve equation. The video enhancement method as described in claim 1, wherein the step of segmenting the video frame into the at least one facial block comprises: segmenting the video frame according to at least one facial feature point in the video frame, wherein the facial block includes at least one facial feature point. The video enhancement method as described in claim 3, wherein the at least one face block includes multiple face blocks, and a boundary of one of the face blocks is adjacent to a boundary of another of the face blocks. The video enhancement method as described in claim 3, wherein at least one facial feature point in the facial block corresponds to an organ in the face. The video enhancement method as described in claim 1, wherein after the step of adjusting the adjusted color level value of the at least one pixel, further comprises: adjusting the color level values of a plurality of adjacent pixels at the boundary between the two face blocks to reduce the difference between the color level values of the adjacent pixels. The video enhancement method as described in claim 6, wherein the step of adjusting the color scale values of the adjacent pixels at the junction between the two facial blocks includes: assigning a distance weight and a block weight to the adjacent pixels respectively, wherein the distance weight is inversely proportional to the distance from the junction, and the block weight is related to the importance of the organ of the facial block to which it belongs; and determining the color scale values of the adjacent pixels according to the distance weight and the block weight of the adjacent pixels. An image processing device includes: a memory for storing a program code; and a processor, coupled to the memory, configured to load and execute the program code to: segment an image frame of a video into at least one face block, wherein the at least one face block corresponds to a face captured by the image frame; adjust the brightness value of at least one pixel in each face block according to a conversion curve, wherein an initial brightness value of the pixel corresponds to an adjusted brightness value in the conversion curve, The conversion curve is related to a polynomial formed according to the initial brightness value and the adjusted brightness value, and adjusting the brightness value of the at least one pixel in each of the face blocks according to the conversion curve includes: substituting an initial color scale value of each of the three primary colors of one of the at least one pixel in each of the face blocks into the polynomial to obtain the adjusted color scale values of the three primary colors; and determining an adjusted brightness value according to the adjusted color scale values of the three primary colors, wherein the adjusted color scale value of each of the primary colors to form a color gradation value of the primary color in the adjusted brightness value; and to adjust a primary color ratio of the at least one pixel in the at least one face block to a target ratio to determine the fine-tuned color gradation value of the at least one pixel, wherein the primary color ratio of the at least one pixel is the proportion of the adjusted color gradation value of the primary color to the sum of the adjusted color gradation values of the three primary colors in the face block, and the processor is further configured to: adjust the primary color ratio to a target ratio according to an S function characteristic, The primary color ratio is related to a definition domain of an S function, and the target ratio is related to a value range of the S function, and the S function characteristics include: when the value in the definition domain of the S function is closer to negative infinity, the corresponding value in the value range of the S function tends to zero; and When the value in the definition domain of the S function is closer to positive infinity, the corresponding value in the value range of the S function tends to one, and when the adjusted color scale value of the primary color is closer to a maximum value, the fine-tuned color scale value of the primary color is closer to one. An image processing device as described in claim 8, wherein the polynomial is a quadratic curve equation, and the processor is further configured to: obtain a brightness representative value of the at least one pixel in the face block, wherein the brightness representative value is related to a statistical indicator of the brightness value of the at least one pixel, and an adjustment curve formed by the quadratic curve equation in a two-dimensional rectangular coordinate system passes through the coordinates of the brightness representative value and a target brightness value in the two-dimensional rectangular coordinate system; and obtain the adjusted brightness value of the at least one pixel in each face block according to the color value of the at least one pixel in the face block and the quadratic curve equation. An image processing device as described in claim 8, wherein the processor is further configured to: segment the image frame according to at least one facial feature point in the image frame, wherein one of the facial blocks includes at least one of the facial feature points. An image processing device as described in claim 10, wherein the at least one face block includes multiple face blocks, and two boundaries of one of the face blocks are adjacent to the boundaries of the other two of the face blocks. An image processing device as described in claim 10, wherein at least one facial feature point in the facial block corresponds to an organ in the face. An image processing device as described in claim 8, wherein the processor is further configured to: adjust the color level values of multiple adjacent pixels at the boundary between the two facial blocks to reduce the difference between the color level values of the adjacent pixels. An image processing device as described in claim 13, wherein the processor is further configured to: assign a distance weight and a block weight to the adjacent pixels respectively, wherein the distance weight is inversely proportional to the distance from the boundary, and the block weight is related to the importance of the organ of the facial block to which it belongs; and determine the color value of the adjacent pixels according to the distance weight and the block weight of the adjacent pixels.

Images