CN104484892A - Image background color identification method - Google Patents

Abstract

The embodiment of the present invention discloses an image background color recognition method, which is applied in the computer field, and can avoid the problems of extremely low efficiency and lack of intelligence in the existing image background color recognition of goods that rely on manual recognition. The method includes: extracting the foreground of the original image through a high-pass filter; removing the foreground to obtain the background of the original image; identifying the background color according to the foreground and the background. Embodiments of the present invention are applied to image background color recognition.

Description

A Method for Image Background Color Recognition

technical field

The invention relates to the field of computers, in particular to an image background color recognition method.

Background technique

As humans enter the information age, computers are increasingly used in various fields. The research on pattern recognition based on digital image processing has attracted more and more attention, but there are still a lot of gaps in the application research on the background color recognition of commodities. The role and significance of background color recognition: extract the background component of a given image and recognize its color, and finally output the name of the visual color, such as "red". At present, it mainly relies on manual identification of the background color of the product, which is extremely inefficient and lacks intelligence.

Contents of the invention

An embodiment of the present invention provides an image background color recognition method to solve the problem of low efficiency and lack of intelligence in the existing image background color recognition of commodities by manual recognition.

A first aspect of the present invention provides an image background color recognition method, comprising:

Extract the foreground of the original image through a high-pass filter;

remove the foreground to obtain the background of the original image;

A background color is identified from the foreground and the background.

According to the first aspect, in a first possible implementation manner, the identifying the background color according to the foreground and the background includes:

The average RGB components of the foreground and the background are respectively extracted, and the background color is identified according to a preset ratio relationship of the RGB components and a preset threshold.

According to the first aspect, in a second possible implementation manner, the extracting the foreground of the original image through a high-pass filter includes:

The original image is subjected to RGB channel separation to obtain a black and white image;

Enhancing the edge of the black-and-white image by high-pass filtering to obtain a high-frequency image;

The high-frequency image is augmented to obtain the foreground of the original image.

The image background color recognition method provided by the embodiment of the present invention greatly improves efficiency by using an image processing algorithm to enable a computer to replace manual positioning of signs.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings that are required in the description of the embodiments or the prior art.

FIG. 1 is a schematic flow chart of an image background color recognition method provided by an embodiment of the present invention;

Fig. 2 is a schematic diagram of expansion in the image background color recognition method provided by the embodiment of the present invention;

Fig. 3 is a schematic diagram of filling in the image background color recognition method provided by the embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention.

FIG. 1 is a schematic flowchart of an image background color recognition method provided by an embodiment of the present invention. This method is mainly used to identify the background color of a product logo, such as a clothing logo, and this embodiment only uses the clothing logo for illustration. Shown in Fig. 1 with reference to, this method comprises the following steps:

10. Extract the foreground of the original image through a high-pass filter.

Wherein, the original image refers to the commodity logo to be recognized. The foreground refers to the landmark area in the original image.

The high-pass filter can filter out low-frequency information in the input, and the input it accepts must be a frequency signal, so it is necessary to convert the image from the spatial domain to the frequency domain through a two-dimensional discrete Fourier transform (DFT).

The frequency of an image is an index that characterizes the intensity of the grayscale transformation in the image. The physical meaning of DFT is to change the grayscale distribution function of the image into a frequency distribution function. The inverse transformation of DFT is to change the frequency distribution function of the image into a grayscale distribution function. .

After the two-dimensional discrete Fourier transform is performed on the original image, the frequency domain center translation operation can also be performed on the transformed image, so that the center area becomes the low frequency component of the new image after the translation operation, making the subsequent processing of the frequency domain image easier. Convenient and intuitive.

The low-frequency components in the center of the image can be suppressed by high-pass filtering, and the high-frequency components can be extracted to obtain the edge of the image.

Specifically, a discrete Fourier transform (DFT) is performed first, and then a high-pass filtering operation is performed. The function of DFT is to transform the image from the spatial domain to the frequency domain. The essence is to change the gray distribution function of the image into a frequency distribution function. The obtained frequency information is processed with a high-pass filter. The function of the high-pass filter is to filter out (suppress ) low-frequency components, but high-frequency components pass.

In this scheme, the Cooley-Turkey fast Fourier transform algorithm can be optimized. Its advantages are:

The number of multiplications required by the computer to calculate the discrete Fourier transform can be greatly reduced, especially the more the number of sampling points N to be transformed, the more significant the saving of the calculation amount of the fast Fourier transform algorithm is. In digital image processing, this algorithm can improve efficiency.

20. Remove the foreground to obtain the background of the original image.

The background is the area around the logo in the original image.

The foreground part of the user's interest in the original image is obtained by image enhancement method, and after removing these foreground parts, only the background of the image remains.

Image enhancement refers to enhancing the useful information in the image, emphasizing the overall or local characteristics of the image purposefully, emphasizing some interesting features, expanding the difference between different object features in the image, and suppressing uninteresting features.

The high-pass filter method carried above is an image enhancement method in the frequency domain, which can obtain high-frequency information in the image.

After obtaining the foreground area of the image, since the gray value of the pixels in this area is larger, it is visually brighter, so the average value of the pixel gray value of the entire image can be calculated, and then the pixels whose gray value is greater than the average value can be set. It is a NaN (not a number) type, so that the entire foreground part becomes a NaN type, thereby realizing the removal of the foreground. Because in the subsequent processing, the area of NaN type will not be included in the processing range, which can be regarded as being removed.

30. Identify the background color from the foreground and background.

Further, in step 30, the identifying the background color according to the foreground and the background may include:

The average RGB components of the foreground and background are extracted respectively, and the background color is identified according to the preset ratio relationship of the RGB components and the preset threshold.

Calculates the RGB average of all pixels in the specified area.

In the RGB color space, there is an inevitable relationship between the visual color and the values of the RGB components. The threshold used in this embodiment is exemplified as follows:

Red(r-g)>35&(r-b)>120;

Green(g-b)>33&(g-r)>60;

Blue(b-g)>33&(b-r)>60;

Note: The value range of RGB is 0~255.

Further, in step 10, extracting the foreground of the original image through a high-pass filter may preferably include the following steps:

101. Separating the original image into RGB channels to obtain a black and white image;

RGB channel separation can turn a color image into a black and white image.

102. Enhance the edge of the black and white image by high-pass filtering to obtain a high-frequency image;

The unsmooth transition between the background and the foreground will form a visual outline. In the frequency domain, the edges of the image appear as high frequencies due to the drastic changes in the gray value. These edges can be extracted and strengthened by high-pass filtering.

103. Expand the high-frequency image to obtain a foreground of the original image.

In the high-pass filtering of the previous step, the high-frequency information of the image is obtained, that is, the edge of the foreground. The foreground edge is expanded (expanded) and the closed area is filled to make the foreground image form a whole, thus obtaining the original The foreground of the image. The expansion method uses morphological dilation and filling.

Dilation and filling operations are basic operations in image processing, and usually do not need to be described in detail. For the sake of completeness, the principles are listed here for the reference of the agent.

Swelling is used to fill breaks in the edge of the sign, bridge cracks in the edge, and make the edge of the sign continuous and closed.

$\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \&CirclePlus;</span>\mathrm{<span\; class="notranslate">\; B</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; (</span>\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; B</span>}}<span\; class="notranslate">\; )</span>}_{\mathrm{<span\; class="notranslate">\; z</span>}}<span\; class="notranslate">\; \&cap;</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \&SubsetEqual;</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; \}</span>$

in, As a structural element, the element after A is expanded by B is the set of all displacements z.

Referring to FIG. 2 , FIG. 2 is a schematic diagram of dilation in an image background color recognition method provided by an embodiment of the present invention.

Filling is a goal to be achieved. The method used is an algorithm based on morphological reconstruction. The so-called reconstruction is to process the original image to make it generate a new image. It is a form involving two images and a structural element. Transformation, one of the images is used as a marker, which is called the starting point of the transformation, and the other image is a mask (the mask image is the original image that needs to be reconstructed), which is used to constrain the change process. Structural elements are used to define connectivity. Referring to FIG. 3 , FIG. 3 is a schematic diagram of filling in an image background color recognition method provided by an embodiment of the present invention.

In this embodiment, the high-pass filtering method is used to extract the background of the static image, instead of the conventional method of image binarization combined with image addition and subtraction, the edge information is obtained by high-pass filtering, and the foreground and background are realized by morphological operations. Segmentation avoids incorrect segmentation caused by improper threshold of simple image binarization, especially when the background and foreground colors are close, the small noise in the image is enough to cause incorrect binarization segmentation, and the high-pass filtering method is based on the edge It is not sensitive to noise and more adaptable. In addition, in terms of color recognition, this embodiment, according to the regularization relationship of the RGB components, is directly mapped to the name of the visual color, which is simple and efficient in calculation, and avoids the huge overhead caused by using the classifier to identify the color. Using the classifier to identify the color The classifier needs to be trained, and a large overhead is also required in the subsequent recognition process.

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (3)

1. An image background color identification method is characterized by comprising the following steps:

extracting the foreground of the original image through a high-pass filter;

removing the foreground to obtain the background of the original image;

and identifying the background color according to the foreground and the background.

2. The method of claim 1, wherein identifying a background color from the foreground and the background comprises:

and respectively extracting the average RGB components of the foreground and the background, and identifying the background color according to a preset RGB component proportional relation and a preset threshold value.

3. The method of claim 1, wherein extracting the foreground of the original image through a high-pass filter comprises:

separating RGB channels of the original image to obtain a black-and-white image;

enhancing the edge of the black-and-white image through high-pass filtering to obtain a high-frequency image;

and expanding the high-frequency image to obtain the foreground of the original image.