AI-based color perception and color tone control method and system

Abstract

The present invention relates to the field of color adjustment, and the present invention relates to a color perception color adjustment control method and system based on AI. The invention obtains a material color adjustment video and extracts a process image, sets a grayscale weight according to the target color and the pigment pixel value before and after color adjustment, grays the process image, and subsequently uses an edge detection algorithm and a grayscale co-occurrence matrix to obtain the edge confusion degree of the stirring process image, calculates the color adjustment completion degree according to the edge confusion degree, and finally determines whether the color adjustment is completed according to the color adjustment completion degree. The present invention customizes the grayscale weight and constructs the edge confusion degree according to the edge features of the image, and significantly improves the production efficiency and quality control capability through automated, precise and efficient technical means, saves resources, and meets environmental protection requirements. It is a better color adjustment control method.

Description

Technical Field

[0001] The present invention relates to the field of color adjustment, and in particular to an AI-based color perception color adjustment control method and system. Background Art

[0002] The color matching process of injection molded products refers to the process of adding masterbatches or pigments to achieve the desired color requirements of plastic products during the production process of plastic products. This process usually includes steps such as adding, stirring, and mixing pigments to ensure that the pigments are evenly distributed in the plastic material. The color matching process has a crucial impact on the appearance and quality of the product, especially in application scenarios that require precise color matching, such as home appliance housings, automotive interiors, etc. The traditional color matching process usually relies on manual experience and visual judgment. In the color matching process of injection molded products, common technical problems mainly include the following aspects: uneven color, long color matching time, difficulty in ensuring accuracy, and lack of automation.

[0003] In order to overcome the technical problems in the traditional color grading process, some image processing and data analysis technologies have been gradually introduced in the prior art to achieve automation and precision of the color grading process. For example, the Chinese patent application document with publication number CN111461996A discloses a method for fast intelligent color grading of images, which uses an intelligent color grading model to intelligently grading the toning image, thereby improving the efficiency of image color grading. However, in these existing technologies, simple color comparison and statistical methods are usually used to determine whether the color adjustment process is completed. When processing complex color adjustment scenes, the accuracy is low and it is easily affected by external factors such as light changes and color occlusion, making it difficult to achieve accurate color detection and color adjustment completion judgment; at the same time, when processing color adjustment images, traditional image processing technology often ignores the detail changes in the image and cannot accurately reflect the dispersion and mixing of pigments during the stirring process, resulting in inaccurate judgment of color uniformity; secondly, the existing technology lacks effective methods to quantify the mixing state of colors, and usually relies on simple color difference calculations, which makes it difficult to accurately reflect the mixing and changes of colors during the color adjustment process, especially when the colors are similar, it is easy to make misjudgments; finally, when processing a large number of color adjustment images, the existing image processing technology has a large amount of calculation and poor real-time performance, making it difficult to meet the requirements of real-time performance and efficiency in the production process. Summary of the invention

[0004] In view of the above problem that it is difficult to accurately reflect the mixing and change of colors in the color adjustment process, in a first aspect, the present invention proposes an AI-based color perception color adjustment control method, comprising:

[0005] The video of the whole process of color adjustment is converted into a group of pictures, and the group of pictures is extracted and preprocessed; during the whole process of color adjustment, pigments are added to materials for color adjustment to obtain the expected target color; the grayscale weights of each color channel are calculated according to the pixel values ​​of the material, pigment and target color, and the grayscale weights are used to characterize the differences between the pixel values ​​of the material, pigment and target color of the corresponding channel; the image is weighted grayscaled according to the grayscale weights to obtain the edge image, and the entropy of the grayscale co-occurrence matrix of the edge image is calculated; the entropy is arranged according to the image shooting order to obtain an entropy sequence; the color adjustment completion is calculated according to the entropy sequence, and the color adjustment completion is calculated as follows:

[0006]

[0007] in, Indicates the degree of color grading completion; represents the mean value of the entropy sequence; represents the maximum value of the entropy sequence; Indicates the minimum value of the entropy sequence; in response to the color adjustment completion being less than a threshold, continues stirring; in response to the color adjustment completion being greater than or equal to a threshold, ends color adjustment.

[0008] The inventors have found that existing image processing technologies usually rely on simple color contrast and statistical methods, which have low accuracy when processing complex color adjustment scenes, and it is difficult to accurately reflect the mixing and change problems of colors during the color adjustment process, making it difficult to achieve accurate color detection and color adjustment completion judgment. The present invention uses AI algorithms and image processing technology to calculate the grayscale co-occurrence matrix and edge confusion degree of the image, which can comprehensively reflect the color mixing state during the color adjustment process, achieve high-precision color adjustment completion judgment, and avoid the limitations of simple color contrast methods. The present invention can quantify the degree of color mixing during the color adjustment process through edge detection and grayscale co-occurrence matrix calculation, and provide a reliable indicator to judge the completion of color adjustment, thereby improving the automation and accuracy of the color adjustment process and ensuring the consistency and quality of product color.

[0009] Furthermore, capturing a video of the entire color grading process and converting it into a set of pictures also includes: using a CCD camera to capture a video of the entire material color grading process, the entire process video including images before color grading, images during the color grading process, and images after the color grading is completed, the image after the color grading is completed being the last frame of the video; using OpenCV to load the entire process video, taking each frame of the video and saving it as a set of pictures.

[0010] Using CCD cameras to shoot improves the clarity and resolution of images, thereby enhancing the ability to capture image details and improving the accuracy of the color grading process. Compared with traditional CMOS cameras, CCD cameras have higher sensitivity and image quality, and are particularly suitable for color grading applications that require high-precision color detection.

[0011] Furthermore, extracting frames and preprocessing the group of images also includes: avoiding saving every frame of the image, selecting frames with a fixed interval of 10s to save, so as to reduce the number of blurred frames; and denoising and sharpening the saved frame images.

[0012] By using a fixed frame extraction method, the quality of image preprocessing is improved, ensuring the accuracy of subsequent grayscale conversion and edge detection. Compared with the simple frame extraction method in the prior art, the image screening step of the present invention can effectively filter out low-quality images caused by motion blur, thereby ensuring the reliability of image data.

[0013] Furthermore, denoising and sharpening the stored frame image further includes: removing image noise using Gaussian blur and sharpening the stored frame image using a Laplace filter.

[0014] Furthermore, the grayscale weight of each color channel is calculated, which also includes:

[0015]

[0016] in, Represents the weight of the r channel; The r channel value representing the pixel value of the material; The r channel value representing the pigment pixel value; The r channel value representing the target color pixel value; Represents the total difference between material and pigment pixel values; Indicates the total difference between the material and target color pixel values; Represents the total difference between the pigment and target color pixel values; obtains the weights of the g channel and b channel respectively according to the above r channel weight calculation method.

[0017] By using a specific formula to calculate the weight of each color channel, it is ensured that when the RGB image is converted to a grayscale image, the grayscale value can accurately reflect the actual situation of each color area before and after color adjustment, thereby improving the accuracy of grayscale processing. Traditional grayscale methods usually use fixed weights and cannot adapt to the needs of different color scenes. The dynamic weight calculation method of the present invention can be adjusted according to actual color changes, thereby enhancing the flexibility and adaptability of image processing.

[0018] Furthermore, obtaining the edge image also includes: using the Canny edge detection algorithm to obtain the edge image on the image after weighted grayscale processing.

[0019] Further, calculating the entropy of the gray level co-occurrence matrix of the edge image also includes: obtaining the gray level co-occurrence matrix of the edge image, and calculating the entropy of the gray level co-occurrence matrix, and the calculation process is as follows:

[0020]

[0021] in, Indicates the degree of edge confusion in the edge image, Represents the element value of the sth row and tth column in the grayscale matrix, Represents the sum of all element values ​​in the gray-level co-occurrence matrix.

[0022] By using the gray-level co-occurrence matrix to calculate the edge confusion formula, the complexity of the image edge can be quantified, providing a reliable indicator to judge the progress of the color adjustment process. The existing technology usually relies on simple color difference calculation, while the present invention introduces edge complexity measurement to more accurately reflect the color mixing state in the color adjustment process, avoiding the problem of misjudgment when the colors are similar.

[0023] Furthermore, in response to the color adjustment completion being less than a threshold, continuing the stirring also includes: setting a color adjustment completion threshold, when the color adjustment completion is less than the set threshold, the color adjustment stirring time is insufficient, continuing the stirring, and reacquiring the color adjustment completion corresponding to the stirring process image and comparing it with the set threshold.

[0024] Furthermore, in response to the color adjustment completion being greater than or equal to a threshold, the color adjustment is ended, and it also includes: when the color adjustment completion is greater than or equal to a set threshold, the color adjustment process is ended, and the color value after the color adjustment is completed is judged whether it is consistent with the target color value. If it does not match, a color adjustment failure signal is issued, and professional personnel add pigments before color adjustment.

[0025] In a second aspect, the present invention provides an AI-based color perception tinting control system, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the AI-based color perception tinting control method of the present invention is implemented.

[0026] The technical effects of the present invention are:

[0027] The present invention customizes the grayscale weights and constructs the edge confusion degree according to the image edge features. Through automated, precise and efficient technical means, it significantly improves production efficiency and quality control capabilities, saves resources, and meets environmental protection requirements. It is a better color control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0029] Figure 1 is a flowchart schematically illustrating a color perception and color adjustment control method based on AI in an embodiment of the present invention;

[0030] Figure 2 is a schematic diagram schematically showing the grayscale during the color mixing and stirring process and after the stirring is completed in an embodiment of the present invention;

[0031] Figure 3 is a schematic diagram schematically showing a logical structure of judging whether color adjustment is completed according to the color adjustment completion degree in an embodiment of the present invention;

[0032] Figure 4 It is a block diagram schematically showing the structure of an AI-based color perception and color adjustment control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0034] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0035] Embodiment of the color perception and color adjustment control method based on AI:

[0036] like Figure 1 As shown, the AI-based color perception color adjustment control method of the present invention includes:

[0037] S1. Obtain material color grading video; obtain color grading process image after preprocessing.

[0038] You can use a CCD camera to shoot a video of the entire process of material coloring. The video of the entire process includes images before coloring, images during coloring, and images after coloring. The image before coloring refers to the image before the pigment is put into the material to be colored but before stirring and mixing. Use OpenCV to load the video, take each frame of the video, and finally save the frame as an image file.

[0039] Using CCD cameras to shoot improves the clarity and resolution of images, thereby enhancing the ability to capture image details and improving the accuracy of the color grading process. Compared with traditional CMOS cameras, CCD cameras have higher sensitivity and image quality, and are particularly suitable for color grading applications that require high-precision color detection.

[0040] Avoid saving every frame, and choose to save frames at a fixed interval of 10s to reduce the number of blurred frames. Use Gaussian blur to remove image noise and use Laplace filter to sharpen the saved frame image to obtain the pre-processed color process image.

[0041] By using the fixed interval frame extraction method, the quality of image preprocessing is improved, and the accuracy of subsequent graying and edge detection is ensured. Compared with the simple frame extraction method in the prior art, the image screening step of the present invention can effectively filter out low-quality images caused by motion blur, ensuring the reliability of image data.

[0042] S2. Obtain the grayscale weight of each color channel; obtain the edge image of the color adjustment process.

[0043] When adjusting the color of injection molding materials, pigments need to be added to the materials. After stirring, it is determined whether the color meets the requirements. If the color of the pigment after stirring does not meet the target color, more pigments are added for color adjustment. When determining whether the pigment is sufficiently stirred, it can be determined based on the number of edges in the material image taken after stirring. When the stirring is not sufficient, there will be pixel values ​​at each position in the material image that are consistent and there will be no edges where the pigments merge. Figure 2 As shown in the middle left picture; when the pigments are fully stirred, there is no edge of fusion of the pigments in the material image, thus keeping the pixels consistent, as shown in the figure below. Figure 2 Therefore, during the stirring process, the edge of the image can be detected in real time to determine whether the stirring is completed.

[0044] When calculating the edge in the image, the RGB image needs to be converted into a grayscale image. Conventional grayscale image algorithms such as the maximum value method and the average value method cannot reflect the actual situation of each color area before and after color adjustment. In the conversion process of the present invention, the weight of each color channel is determined according to the pixel value of the material before color adjustment, the pixel value of the pigment and the pixel value of the color adjustment target color. In this way, the edge in the grayscale image can be made more obvious according to the color difference of the color channel.

[0045] When an RGB image is converted to a grayscale image, the weight of each color channel is calculated as follows:

[0046]

[0047] in, Represents the weight of the r channel; The r channel value representing the pixel value of the material before color adjustment; The r channel value representing the added pigment pixel value; The r channel value representing the color adjustment target color pixel value; Indicates the total difference in pixel value between the material before color adjustment and the added pigment; Indicates the total difference in pixel value between the material before color grading and the target color grading; Indicates the total difference between the added pigment pixel value and the toning target color pixel value.

[0048] when The larger the value, the greater the difference between the red channel value of the material before color adjustment and the red channel value of the added pigment. The larger the value, the greater the weight of the red channel. Therefore, when graying the image, the red channel value should have a larger proportion. The larger the value, the greater the difference between the red channel value of the material before color adjustment and the red channel value of the target color. The larger the value, the greater the weight of the red channel. Therefore, when graying the image, the red channel value should have a larger proportion. The larger the value, the greater the difference between the red channel value of the added pigment and the red channel value of the color adjustment target color. The larger it is, the greater the weight of the red channel, so when graying the image, the red channel value should have a larger proportion.

[0049] According to the above calculation process, the weights of the g channel and the b channel can also be obtained. According to the difference of each color channel of the three, a grayscale weight is given to each channel. The greater the difference in the pixel values ​​of the three color channels, the greater the weight given to the color channel. In this way, the grayscale difference of each area in the captured image is greater, and the edge obtained in the subsequent edge detection is more accurate and fine.

[0050] By using a specific formula to calculate the weight of each color channel, it is ensured that when the RGB image is converted to a grayscale image, the grayscale value can accurately reflect the actual situation of each color area before and after color adjustment, thereby improving the accuracy of grayscale processing. Traditional grayscale methods usually use fixed weights and cannot adapt to the needs of different color scenes. The dynamic weight calculation method of the present invention can be adjusted according to actual color changes, thereby enhancing the flexibility and adaptability of image processing.

[0051] In another embodiment, the weight of each color channel can also be calculated in other ways. For example, the denominator in the formula can be the sum of the pixel values ​​of all channels of the material, pigment and target color; that is, the grayscale weight in the present invention mainly reflects the differences between the material, pigment and target color in each color channel.

[0052] S3, obtaining a gray-level co-occurrence matrix corresponding to the edge image; and calculating the degree of disorder of the edge of the image.

[0053] The image obtained by S1 is grayed out according to the weight obtained by S2. After graying, the Canny edge detection algorithm is used to obtain the edge image. The edge image can be calculated at any point in the edge image. The value a, in Direction, distance point for point The value of The probability of , Represents the minimum gray value in the edge image. Represents the maximum gray value in the edge image. Traversing all the grayscale values ​​in the edge image requires a lot of calculation, so 8 grayscale levels are set for the edge image.

[0054]

[0055] Where H(i,j) is the grayscale value at the i-th row and j-th column of the edge image, and the symbol [ ] is the rounding down symbol. In the direction The number of occurrences is calculated to obtain the gray-level co-occurrence matrix of the edge image.

[0056] The entropy of the gray-level co-occurrence matrix is ​​calculated, which reflects the degree of confusion at the edge of the image. The calculation formula is as follows:

[0057]

[0058] in, Indicates the degree of edge confusion in the edge image, Represents the element value of the sth row and tth column in the grayscale matrix, It represents the sum of all element values ​​in the gray-level co-occurrence matrix. The greater the edge chaos, the more edges there are in the image at this time, and the lower the possibility of pigment mixing is.

[0059] By using the gray-level co-occurrence matrix to calculate the edge confusion formula, the complexity of the image edge can be quantified, providing a reliable indicator to judge the progress of the color adjustment process. The existing technology usually relies on simple color difference calculation, while the present invention introduces edge complexity measurement to more accurately reflect the color mixing state in the color adjustment process, avoiding the problem of misjudgment when the colors are similar.

[0060] S4. Obtain an edge feature sequence of the real-time captured image according to the degree of confusion and calculate a color adjustment completion degree; and determine whether the color adjustment is completed according to the color adjustment completion degree.

[0061] According to the method for obtaining the color-adjusting process image after preprocessing described in S1, the whole process image of real-time shooting and color-adjusting is obtained. According to the method for calculating the degree of disorder of edge image described in S3, the degree of disorder of edge in each image is calculated, and the edge feature sequence of the image is obtained according to the time sequence of the image. The edge feature sequence is composed of the degree of disorder of edge of the image. For example, 10 color-adjusting process images are obtained in total, and the degree of disorder of edge of each image is calculated as follows: , then the edge feature sequence of the color adjustment process is .

[0062] Whether the color adjustment is completed is determined based on the eigenvalue and feature stability of the edge feature sequence. The calculation formula is as follows:

[0063]

[0064] in, Indicates the degree of color grading completion; represents the mean value of the entropy sequence in the toned image; represents the maximum value of the entropy sequence in the toned image; Indicates the minimum value of the entropy sequence in the color-corrected image within the set range.

[0065] when The larger the value, the greater the confusion of the image in the color adjustment process. The smaller it is, the lower the degree of color mixing is, indicating that the pigment is not fully stirred; The larger the value, the greater the extreme difference in the degree of chaos of the image in the color adjustment process. The smaller it is, the less complete the color mixing is, which means the paint is not stirred sufficiently.

[0066] The present invention sets the color adjustment completion threshold ,like Figure 3 As shown, when the coloring completion is less than the set threshold, the coloring stirring time is insufficient, and the stirring is continued and the coloring completeness is recalculated and compared with the threshold; when the coloring completion is greater than or equal to the set threshold, the coloring is completed, and then it is determined whether the color value after the coloring is completed is consistent with the target color value. If it does not match, a coloring failure signal is issued, and professionals are required to add pigments before coloring; if it matches the target color value, the coloring is completed. Set as Experience Value , implementers can make adjustments according to actual conditions to meet color requirements.

[0067] When the color matching is not completed, a color matching failure signal is issued, and professionals add pigments before color matching, which improves the accuracy of the color matching process and the color consistency of the final product, and reduces the scrap rate. Color matching failure in the prior art usually relies on manual discovery and adjustment, which is inefficient and prone to errors. The present invention can detect problems in time and make corrections through automatic detection and prompts, ensuring the efficiency and stability of the color matching process.

[0068] AI-based color perception color grading control system implementation example:

[0069] On the other hand, the present invention also provides a color perception and color adjustment control system based on AI. Figure 4 As shown, the AI-based color perception and tinting control system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the AI-based color perception and tinting control method according to the first aspect of the present invention is implemented.

[0070] The AI-based color perception color grading control system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface, whose settings and functions are known in the art and will not be described in detail here.

[0071] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0072] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0073] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. The AI-based color perception color grading control method is characterized by: include: The video of the whole process of color adjustment is shot and converted into a group of pictures, and the group of pictures is framed and pre-processed; during the whole process of color adjustment, pigment is added to the material to adjust the color to obtain the expected target color; The grayscale weight of each color channel is calculated according to the pixel values ​​of the material, pigment and target color. The grayscale weight is used to characterize the difference between the pixel values ​​of the material, pigment and target color of the corresponding channel. The calculation method is: ； in, Represents the weight of the r channel; The r channel value representing the pixel value of the material; The r channel value representing the pigment pixel value; The r channel value representing the target color pixel value; Represents the total difference between material and pigment pixel values; Indicates the total difference between the material and target color pixel values; Represents the total difference between the pigment and target color pixel values; obtains the weights of the g channel and the b channel respectively according to the above r channel weight calculation method; Performing weighted grayscale processing on the image according to the grayscale weight, obtaining an edge image, and calculating the entropy of the grayscale co-occurrence matrix of the edge image; Arrange the entropies according to the image shooting order to obtain an entropy sequence; calculate the color adjustment completion degree according to the entropy sequence, and the color adjustment completion degree is calculated as follows: ； in, Indicates the degree of color grading completion; represents the mean value of the entropy sequence; represents the maximum value of the entropy sequence; represents the minimum value of the entropy sequence; In response to the color adjustment completion degree being less than a threshold value, the stirring is continued; in response to the color adjustment completion degree being greater than or equal to the threshold value, the color adjustment is ended.

2. The AI-based color perception color grading control method according to claim 1, characterized in that: The video of the entire color grading process is converted into a set of pictures, including: A CCD camera is used to shoot a video of the entire process of material color grading, wherein the entire process video includes images before color grading, images during color grading, and images after color grading is completed, wherein the image after color grading is the last frame of the video; OpenCV is used to load the entire process video, and each frame of the video is taken and saved as a group of images.

3. The AI-based color perception color grading control method according to claim 1, characterized in that: Extracting frames from the group of images and performing preprocessing, including: Avoid saving every frame image, and select frames with a fixed interval of 10s to save to reduce the number of blurred frames; denoise and sharpen the saved frame images.

4. The AI-based color perception color grading control method according to claim 3, characterized in that: De-noise and sharpen the saved frame image, including: The saved frame images are then subjected to Gaussian blur to remove image noise and Laplacian filter for sharpening.

5. The AI-based color perception color grading control method according to claim 1, characterized in that: Get edge images, including: The Canny edge detection algorithm is used to obtain the edge image of the weighted grayscale image.

6. The AI-based color perception color grading control method according to claim 1, characterized in that: Calculating the entropy of the gray level co-occurrence matrix of the edge image includes: Obtain the gray level co-occurrence matrix of the edge image, and calculate the entropy of the gray level co-occurrence matrix. The calculation process is as follows: ； in, Indicates the degree of edge confusion in the edge image, Represents the element value of the sth row and tth column in the grayscale matrix, Represents the sum of all element values ​​in the gray-level co-occurrence matrix.

7. The AI-based color perception and color grading control method according to claim 1, characterized in that: In response to the color mixing completion degree being less than a threshold, continuing to stir, including: A color adjustment completion threshold is set. When the color adjustment completion is less than the set threshold, the color adjustment stirring time is insufficient, and stirring is continued. The color adjustment completion corresponding to the stirring process image is re-obtained and compared with the set threshold.

8. The AI-based color perception color grading control method according to claim 1, characterized in that: In response to the color adjustment completion degree being greater than or equal to a threshold, ending the color adjustment, including: When the color adjustment completion degree is greater than or equal to the set threshold, the color adjustment process ends. The color value after the color adjustment is completed is judged whether it matches the target color value. If it does not match, a color adjustment failure signal is issued, and professionals will add pigments before color adjustment.

9. AI-based color perception and color grading control system, characterized in that: It includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the AI-based color perception and color adjustment control method described in any one of claims 1 to 8 is implemented.

Images