Dark-channel-prior-based quick defogging method of traffic monitoring image

Abstract

The invention discloses a dark-channel-prior-based quick defogging method of a traffic monitoring image. The morphological closing operation and Gaussian smoothing filtering are used for replacing guiding filtering or bilateral filtering, so that local smoothness of the transmittance of a traffic monitoring image at the same depth of field is ensured with the low computational complexity. Meanwhile, with an average RGB value of a high-brightness pixel set and upper and low boundary constraints of an atmospheric light value, a phenomenon of unreasonable atmospheric light value estimation because of various interference factors during the traffic monitoring process is avoided. Besides, the contrast of the output image is enhanced based on nonlinear transformation of the transmittance, so that an objective of realizing defogging of a traffic monitoring image quickly and effectively is achieved.

Description

technical field [0001] The invention relates to the field of traffic monitoring image processing, in particular to a traffic monitoring image defogging method with fast operation speed and good defogging effect based on dark channel prior. Background technique [0002] The traffic monitoring system is a subsystem of intelligent transportation, which establishes a fast communication link between roads, vehicles and drivers, and the role that the traffic monitoring system can play depends largely on the weather conditions. Under severe weather conditions such as fog and haze, the video or image captured by traffic monitoring equipment will have problems such as decreased contrast, decreased clarity, and color distortion, which will reduce the use value of the video or image and affect the effectiveness of traffic monitoring. Removing or reducing the impact of dust and haze particles and improving the quality of monitoring images is one of the key technical links to improve the...

AI Technical Summary

Problems solved by technology

Although these methods can perform ideal processing on a single image, especially for images with a lot of depth of field, the processing effect has been greatly improved, but Han Zhengting and others believe that in general, the transmittance in the local area It should be constant. Guided filtering on foggy images will make the details of the same depth of field too rich (especially for long-distance wide-angle images such as traffic monitoring images), which is not in line with natural laws. The guide map of the atmospheric light curtain conducts guide filtering on the atmospheric transmittance, and then combines the atmospheric physical scattering model to complete the defogging

However, the computational complexity of this algorithm is still high, which cannot well meet the real-time requirements of traffic monitoring image processing.

Images