Video Image Enhancement Technique For Low Illumination Based On Spatial-temporal Model

In order to observe the trend of development of things,people usually use the video which with large information capacity to record or obtain information.At night and low light scenes,video captured by non-infrared camera will appear low brightness,noise and low contrast phenomenon,and the infrared camera video can't show the color information of the images,as a result,the videos which captured by the existed cameras can't up to the requirements for the particular application.In order to obtain high-quality color video images adapted to the human eye observation,this paper proposes an illumination enhancement algorithm based on Spatio-Temporal model for non-infrared video images.The proposed algorithm is divided into three parts: filtering denoising,detail enhancement and brightness enhancement.In order to increase the amount of reference information of low illumination video images,this paper combines the adjacent frame information and spatial neighborhood information to form a new reference pixel set.In the area of noise removal,the clustering method is used to classify the reference pixels and evaluate the pixel weights of each class.Then,the weights of each class of pixels are fused to the bilateral filtering algorithm for image denoising.In the luminance enhancement,the brightness is segmented according to the low point of reference set of pixels' distribution and improved by tone mapping method.In the detail enhancement,this paper combines the edge extraction operator to locate the details of the image,and mapping the details of the image according to the neighborhood distribution of the details.In this paper,the classical algorithm of image enhancement is improved and combined.The experimental results show that the algorithm is suitable for the texture of low illumination or uneven illumination video images.The proposed algorithm improves the luminance enhancement effect of the previous low illumination video image algorithm,and keeps the details of the image well while removing the noise of the image.