Image visual effect enhancement method based on high and low frequency separation

Abstract

The invention relates to an image visual effect enhancement method based on high-low frequency separation, which comprises the following steps: S1, collecting images, and forming an original image setby all the collected high-definition images; S2, performing post-processing on the original image to enhance the visual effect, and obtaining a post-processed image set; S3, smoothing the original image and the post-processing image by adopting a Gaussian filter, and extracting low-frequency components of the original image and the post-processing image; S4, establishing a context aggregation network model based on full convolution, inputting low-frequency components of the original image and the post-processing image as image sequences into a network for training, and updating parameters inthe context aggregation network model to obtain a trained network model; and S5, superposing the high-frequency component of the to-be-processed image and the trained low-frequency component processedby the network model to obtain a visual effect enhancement image of the to-be-processed image. The method is simple and effective, details in the original image can be reserved as much as possible, and the finally obtained visual effect enhancement image is better in effect.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for enhancing the visual effect of an image based on high and low frequency separation. Background technique [0002] With the continuous development of economy and society, the continuous advancement of science and technology, and the continuous reduction of the cost of image acquisition and storage, using cameras to record beautiful moments in life has become an important way for people to preserve memories. Digital image and its processing have gradually penetrated into every corner of our life and work. Image enhancement is an important branch of digital image processing, and its purpose is mainly to improve the visual effect of the image for a given image application. [0003] At present, there are many methods commonly used to enhance the effect of images, such as traditional double-pass filter operators, Gaussian filter operators, etc., but they can only ...

AI Technical Summary

Problems solved by technology

[0005] In the process of visual enhancement of ultra-clear professional portrait photos, we found that if the training images are directly end-to-end, the details of the images may not be well preserved, that is, the edge information of the images is easy to be lost, resulting in some images with poor light and dark transitions. It is a natural phenomenon. Therefore, although the above methods can also enhance the visual effect of the image, they all have a disadvantage, that is, the algorithm model will cause the image to lose some details, such as the pores on the face, etc., making professional portrait photos lack texture.

Through the evaluation of professional retouchers and the statistical evaluation of experimental results, it is found that many pictures will have color casts and the skin care effect will cause problems such as unnatural light and dark transitions and lack of texture, and even some photos will have color casts.

Images