Detail Maintained Low-Light Video Image Enhancement Algorithm

With the development of modern society, people need to acquire nighttime images under low illumination conditions in more and more areas. However, such images are usually not easily observed by the human eyes, and therefore it is necessary to enhance nighttime images. In order to enhance the effect of nighttime image and increase the reference information content of low-illumination video images, this paper combines temporal neighboring information and spatial neighboring information, and takes two adjacent frames before and after the current frame to form a new reference pixel set. The clustering method is used to classify the reference pixel set in removing noise, and the weight of each type of pixel is determined according to the similarity and the proportion of pixel gray values, and then this weight is integrated into the bilateral filter algorithm to denoise the image. Experiments show that the proposed algorithm is suitable for textured low-light or uneven-light video images. This algorithm improves the brightness enhancement effect of the previous low-illumination video image algorithm and preserves the detail information of the image while removing the noise of the image.