Ultra-low light imaging method based on multi-granularity cooperative network

Abstract

The invention belongs to the field of image processing and computer vision, and discloses an ultra-low light imaging method based on a multi-granularity cooperative network, and the method comprises the following steps: S1, collecting original signal data of a camera, and carrying out the black level correction; s2, carrying out gain processing on the signal after black level correction, and thencarrying out dimension reduction processing on the signal data after gain; and S3, inputting the signal data subjected to dimension reduction processing into a multi-granularity cooperative neural network, and converting the multi-granularity cooperative neural network into an sRGB space in a learning mode, the multi-granularity cooperative neural network comprising a plurality of single-granularity networks connected in sequence, and the last single-granularity network being a twin network of the first single-granularity network. According to the invention, higher peak signal-to-noise ratio (PSNR) and structural similarity measurement (SSIM) are realized, and a better visual effect is achieved.

Description

technical field [0001] The invention belongs to the field of image processing and computer vision, and in particular relates to an extremely low-light imaging method based on a multi-granularity cooperation network. Background technique [0002] Imaging under low-light or extremely low-light conditions has always been a very difficult task. Imaging devices have low signal-to-noise ratios under low-light or extremely low-light conditions. Traditional image signal processing (ISP) algorithms process The image will have problems such as noise, blur, color distortion, etc. To solve this problem, one strategy is to increase the exposure time to obtain a sharp image, but due to camera shake or object movement, the increase in exposure time will cause blurring, so this strategy is not suitable for video shooting. Another tactic is to turn on the flash, but this tactic can make the image look unnatural. [0003] Until now, many methods have been proposed to enhance the quality of ...

AI Technical Summary

Problems solved by technology

However, the traditional nighttime image enhancement technology mainly faces two problems in practical applications: 1. It usually leads to overexposure or underexposure in some areas of the enhanced image, and at the same time produces a lot of noise, chromatic aberration and other problems; 2. In extremely weak light environment, especially in terms of protecting high dynamic range (HDR), it is still difficult to obtain satisfactory results

Images