Image processing method, device and equipment and storage medium

Abstract

The invention discloses an image processing method, device and equipment and a storage medium. The method comprises the following steps: extracting an image frame in a video image as an initial illumination image; smoothing texture details of the initial illumination image; and improving the edge precision of the initial illumination image after smoothing processing to serve as a target image. According to the invention, an image frame in a video image is extracted as an initial illumination image; smoothing processing is performed on texture details of the initial illumination image; the image quality is improved, the edge precision is improved, the output target image is enabled to be smooth with excessive texture details, more contour information in the image is reserved, the display effect is improved, rapid processing of the enhanced image in dark light is realized, the time overhead of an image processing process is reduced, the real-time performance of video recording is improved, and the user experience degree is enhanced.

Description

Technical field [0001] The embodiments of the present invention relate to the technical field of image processing, and in particular to an image processing method, device, device, and storage medium. Background technique [0002] With the development of Internet technology, online video services have become the focus of current research. However, due to the impact of the video recording environment, the definition of video images is often low due to the limitation of light conditions during recording. In order to improve the video recording effect, often Use dark light enhancement technology to process the image to improve the video recording effect. [0003] In the prior art, the dark light enhancement method of Retinex theory is currently a widely used image enhancement method. The method decomposes the input image into the product of the light map and the material map, then brightens the light map, and finally enhances The resulting light map is multiplied by the material map t...

AI Technical Summary

Problems solved by technology

[0002] With the development of Internet technology, online video services have become the focus of current research. However, due to the influence of the video recording environment, video images are often recorded with low video clarity due to light conditions. In order to improve the video recording effect, often Image processing with dark light enhancement technology to improve video recording effect

[0003] In the existing technology, the dark light enhancement method of Retinex theory is a widely used image enhancement method at present. This method decomposes the input image into the product of the light map and the material map, then brightens the light map, and finally enhances The final light map is multiplied by the material map to obtain the output image. However, since the accurate solution of the light map involves large-scale sparse matrix operations, the dark light enhancement technology based on the Retinex theory takes a lot of time and requires high real-time performance during video recording. Adopting the Retinex theory does not apply

Images