Dark channel and image segmentation-based traffic image defogging method and system

Abstract

The invention discloses a dark channel and image segmentation-based traffic image defogging method and system. The method comprises the steps of segmenting a foggy traffic image to obtain a close-shot region and a sky region of the foggy traffic image; taking an average intensity value of the foggy traffic image as an atmospheric light coefficient, and calculating atmospheric light transmissivity of a scene in combination with a dark channel of the foggy traffic image; and performing adaptive correction on the atmospheric light transmissivity of the sky region in the atmospheric light transmissivity of the scene, and obtaining a defogged traffic image in combination with the atmospheric light coefficient and a degradation model of the foggy traffic image. According to the method and the system, the traffic image is defogged by comprehensively adopting the image segmentation and dark channel methods, and the condition that the transmissivity of the sky region after defogging is relatively low or is a negative value is avoided through the adaptive correction, so that the effectiveness and universality are higher; and the average intensity value of the sky region of the foggy traffic image is taken as the atmospheric light coefficient, so that the rationality is higher. The method and the system can be widely applied to the field of image processing.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a traffic image defogging method and system based on dark channels and image segmentation. Background technique [0002] In recent years, with the advancement of urbanization and the rapid development of industrialization in various parts of China, environmental problems have become increasingly serious. At the same time, hazy weather has also appeared more and more frequently, which has seriously affected the accuracy of image information obtained by the monitoring system. Taking the highway monitoring system as an example, due to the dense fog, the visibility of the road is greatly reduced, and the road condition information obtained by the driver through vision is often inaccurate, which affects the correct judgment of the environment and easily leads to traffic accidents. At the same time, the degraded images obtained in foggy days also caused great difficulties in monitoring ...

AI Technical Summary

Problems solved by technology

[0005] 1) Based on the dark channel prior theory, for images with bright areas such as sky and white clouds that do not have dark channel, it is easy to calculate the transmittance of the sky area due to the use of the dark channel prior theory Low (because the dark channel prior theory approximates the depth of field in the sky area to be infinite), or even negative (because the dark channel prior theory estimates the atmospheric light coefficient smaller than the real value), it is difficult to achieve effective dehazing of the image , not suitable for dealing with images containing large sky areas such as traffic images, and has poor versatility

[0006] 2) Usually, the first 0.1% of the pixels are extracted from the dark channel image of the foggy image according to the brightness, and then the maximum value corresponding to these pixels in the original image is taken as the atmospheric light coefficient. In this way, the atmospheric light of each channel will appear When the coefficients are all close to the maximum pixel value of 255, local color spots and color casts will appear in the image after defogging, which is not reasonable enough

Images