Low-illumination color image real-time enhancement method and system

Abstract

The invention discloses a low-illumination color image real-time enhancement method and system. The system comprises an RGB-to-YCbCr module which is used for converting a fogged image from an RGB color space to a YCbCr color space, and obtaining a brightness map of the image, a maximum / minimum filtering module which is used for obtaining an atmospheric light value and rough and fine dark channel graphs, a dark channel image synthesis module which is used for manufacturing a mask according to the brightness image and synthesizing the rough dark channel image and the fine dark channel image intoa colorful and natural dark channel image according to the mask, and a defogging module which is used for calculating the image transmissivity according to the dark channel graph and defogging according to the dark channel prior principle. According to the method, the characteristics that the enhanced image is natural in color and the image is not distorted can be kept.

Description

technical field [0001] The invention belongs to image enhancement technology, in particular to a low-illuminance color image real-time enhancement method and system. Background technique [0002] In the nighttime environment, the light is insufficient and the light source is single, resulting in images acquired by general image acquisition equipment generally having blurred images, too low brightness, insufficient contrast, serious loss of object detail information, and color distortion. The information is very limited, which brings difficulties for many applications. For example, the driving recorder, night weather monitoring and driverless night driving system cannot function normally. Therefore, it is becoming more and more important to study how to quickly obtain clear images at night on the edge side. People from various countries have also carried out corresponding research. These methods are mainly divided into the following categories: [0003] 1. Histogram equaliz...

AI Technical Summary

Problems solved by technology

Suppressing the low-frequency components reduces the dynamic range of the image; increasing the high-frequency components increases the contrast, but this method has poor visual effects on low-light images, and image details will be lost. distortion

[0005] 3. Enhancement algorithm based on Retinex model: This algorithm can enhance contrast enhancement, compress dynamic range, and have good color constancy, but it will amplify image noise, appear halo artifacts, and appear in particularly dark areas of the image "Mosaic" phenomenon, and the algorithm is cumbersome

[0006] 4. Dark channel defogging theory algorithm: This type of method can achieve better image enhancement effect, is not easy to introduce noise, and has better color constancy, but the image enhanced by this method is not easy to control the parameter size in the defogging theory , there may be improper image enhancement, color distortion and obvious artificial processing traces. When using soft matting method to optimize the transmittance, the calculation amount is large and the time complexity is high

[0007] In summary, the existing low-light image enhancement algorithms will cause color distortion, halo, and excessive contrast enhancement to the enhanced image while enhancing the image, and some algorithms are time-consuming, cumbersome to program, and the algorithm is not real-time

Images