Research On Night Fog Image Enhancement Based On Improved Dark Channel Algorithm

With the continuous improvement of people's living standard,the demand for image collection and acquisition is growing Both the ordinary color images obtained in the daytime and the infrared images obtained in the low-illumination environment at night all have received a lot of attention and research.Infrared image is an image formed by measuring the thermal radiation difference between the target itself and the environment of obtaining the thermal infrared images with different intensities.Infrared cameras,cameras and other infrared image acquisition equipment have been widely used in security monitoring,traffic guidance,wildlife observation,remote sensing imaging television guidance and other aspects.However,in practical application,infrared image acquisition equipment is prone to be interfered by various external environmental factors,resulting in a decline in the quality of acquired infrared images,which seriously affects the identification and determination of targets in infrared images,especially the interference of fog.The defogging of infrared image will directly affect the subsequent image processing at the higher level in the observation system.Therefore,the research on the defogging of infrared image has important practical significance and value for the improvement and application of monitoring system for road traffic security and wildlife trackingIf the fog is regarded as a kind of noise,the processing of the denoising(defogging)of image,will make the foggy image to the non-foggy image,then the standard of defogging is obviously very objective.Image enhancement and image restoration are two main methods of image defogging at present.The former emphasizes objective standards,while the latter emphasizes subjective standards.However,there are some technical overlaps between them,which ultimately achieve the goal of improving image quality.Image defogging is the most typical example of these two technologies intersecting with each other.The existing defogging methods have –good defogging effect when dealing with ordinary color image with fog,but not in the processing effect of infrared image in night fog environment.The existing methods of defogging cannot keep the complete details of the edge texture when removing the noise of the infrared image in the night fog environment,that is,excessive smoothness will cause the loss of details,and the noise will not be easily eliminated if the details are to be maintained.To solve the above problems,this paper decomposes the infrared image based on the algorithm for existing image defogging,and then makes corresponding enhancement optimization according to the characteristics of the structure layer and texture layer of the decomposed infrared image.Firstly,this paper proposes a solution to improve the dark channel algorithm for the halo phenomenon in texture layer processing.By adding a standard deviation weighting factor,the guidance filter in the dark channel algorithm is restricted to avoid the halo phenomenon when processing the edge,texture and other details in the infrared image texture image.The detail layer and texture part of the infrared image texture image are more fine and clear.Finally,the processed infrared image texture layer is combined with the infrared image structure layer processed by Retinex algorithm for weighted fusion,with the fused infrared image,the purpose of fog removal and quality reduction restoration can be realized,the infrared image can remove the influence of fog and retain the texture edge and other details in the image as much as possible,then enhance the evaluation quality of subjective visual and objective effect.