Self-adaptive image toning method and system, storage medium and electronic equipment

Abstract

The invention relates to the field of picture processing, in particular to a self-adaptive image toning method and system, a storage medium and electronic equipment. The method comprises the following steps: 1, acquiring a to-be-processed image, a source image and a target image; 2, preprocessing the source image and the target image to obtain the source image and the target image after format conversion; step 3, training a BP neural network model through the source image after format conversion and the target image; step 4, inputting the to-be-processed image into a BP neural network model to obtain an output result; and 5, performing white balance processing on the output result to obtain a final image. According to the method, the effect of coordinating the RGB channels can be achieved.

Description

technical field [0001] The present invention relates to the field of image processing, in particular to an adaptive image color correction method, system, storage medium and electronic equipment. Background technique [0002] The storage and presentation form of a general color image in a computer can be expressed as a tensor of dimension H×W×3. H and W are image height and width respectively, and 3 represents RGB (red, green and blue) three channels. The RGB ternary value of each pixel determines the color presented by the point. To realize the color adjustment algorithm of the image, it is essentially to transform the pixel values ​​of the whole image. The adjustments of brightness, contrast, saturation, etc. have corresponding rules for processing image pixel values. As for various filters, the same is true but the rules are more complicated. Therefore, image toning is logically achievable and operable. [0003] For the GUI interface provided by the app, the user can ...

AI Technical Summary

Problems solved by technology

[0003] For the GUI interface provided by the app, users can only manually test the image color, and need to try a combination of various indicators (brightness, contrast, etc.) transformations. If there is no relevant experience and knowledge, it may be difficult to achieve the expected results

In addition, manual color grading can only be used for a small number of pictures, and the algorithm cannot be extended to batch operations

And there is more redundancy in the degree of freedom in the prior art

In addition, the parameter learning process of the BP neural network tends to fall into local extremums, resulting in difficulty in convergence; because the parameter optimization of the three channels is independent to a certain extent, the RGB three channels of the output result may be inconsistent, resulting in a warmer overall tone of the image or cold

Images