Low-light Image Ehancement And The Application In Video Surveillance Research

Video surveillance as an extension of the human visual function,which has a wide range of applications,such as the road traffic,electric power,national security,public places and many other fields.But in the actual video surveillance system,due to bad weather or low-light environment,such as cloudy,uneven illumination or night,which causes a decline in image quality,lower image contrast,fewer image detail,as a result it will affect the recognition of the further work.In order to make full use of low-light video,it needs to be image sharpening,that is to be enhanced.At present,the commonly used image enhancement algorithms have histogram equalization algorithm,image fusion algorithm,and the Retinex algorithm,etc.The paper mainly aims to enhance the low-light video surveillance image,its main work as follows:Firstly,the paper proposes an improved Retinex low-light image enhancement algorithm based on gradient domain guided image filtering.It use the gradient domain guided image filtering instead of gaussian filtering to extract image’s illumination component,and obtain reflection component,then enhance them by using nonlinear Sigmoid and Gamma transform,separately.At last take antilog of the sum of two components to get the enhanced image.The algorithm can eliminate the the "halo" and color distortion phenomenon which appeared in the traditional Retinex algorithm.Secondly,the paper puts forward an image fusion enhancement algorithm based on single low-light image.The algorithm suits for video surveillance images which has a great difference between front and rear frame image information,not suitable for using multiple frames image fusion enhancement algorithm.The algorithm enhance the low light image contrast with different proportions,and combined into a series of fused image with the original image,then fusion these fused image according to certain criterias,finally get the enhanced image.The algorithm is easy to implement,do not need to multiple source images,it can improve the contrast of the image and get the detail of the image.Finally,I apply the improved Retinex low-light image enhancement algorithm based on gradient domain guided image filtering to the video surveillance system.Design the video image acquisition and video image enhancement application software,then enhance the low-light video images,very low-light video images and highlight video images,through the objective analysis of the experimental results,it is concluded that,the algorithm is suitable for processing low-light video image.