Research On Tone Mapping Scheme For High Dynamic Range Video Based On Temporal Correlation

With the rapid development of multimedia technology,high-definition videos and images have gradually become popular in people's daily life.High Dynamic Range(HDR)image is a type of high-definition image.It has higher dynamic range than normal images,contains richer details and a good overall contrast.Similarly,HDR video can also bring people closer to the feeling of a real natural world.HDR videos have a wide range of applications in many fields,such as military,medical,satellite remote sensing and so on.However,in the daily life,the display device can only display low dynamic range images,cannot display HDR video directly.HDR image needs to be compressed into a Low Dynamic Range(LDR)image to display by a tone mapping algorithm.HDR video is composed of many HDR images.To display HDR video on an ordinary display,the traditional method needs to perform tone mapping processing on each HDR image,compressing its dynamic range,and making it suitable for encoding and transmission.The time complexity is very high.Due to the inter-frame correlation between adjacent images of video,the paper abandons the traditional frame-by-frame tone mapping method and study a new HDR video compression scheme.The key frame extraction method of the color histogram is used to perform key frame-based grouping on the HDR image sequence,and the first frame of each group is set as a reference frame.Only the reference frame is subjected to tone mapping processing.Rest frames of the image in the group are used for block-based motion estimation separately with the reference frame.The motion estimation algorithm uses an adaptive rood pattern search method and a mean absolute deviation matching criterion.Simulation results show that under the condition that the block size is 100?100 and the similarity threshold is 20,compared with the traditional method,the proposed compression algorithm can maintain the structural similarity of 89% and the processing speed about 2.7 times.In the case of a fixed matching similarity threshold,as the block size increases,the image quality obtained by the compression algorithm decreases,but the time required for processing decreases.In the case of a fixed block size,as the similarity threshold increases,the image quality decreases slightly,but the processing time decreases accordingly,so the selected block size and the matching threshold need to be compromised.In the proposed HDR video compression scheme,it is very important to select a proper tone mapping algorithm.This paper adopts a tone mapping algorithm based on bilateral filtering and analyzes three parameters of spatial neighborhood standard deviation,pixel brightness standard deviation and contrast compression parameters.The simulation results show that the two values of the spatial standard deviation have little effect on the result,and the pixel brightness standard deviation of 0.7 has a better result image.An adaptive contrast compression method is studied.The logarithm function used in the compression of the base layer can adjust the base according to the actual brightness value of the image,so that the compression ratio is automatically adjusted in the dark area and the highlight area.The simulation results show that the light and dark levels of the image obtained by this method change more obviously and the visual effect is better.For the unprocessed detail layer,different compensation measures are taken according to the brightness changes before and after the compression of the base layer.Combines YUV and HSV color spaces to optimize color retention.The simulation results show that the optimized bilateral filtering tone mapping algorithm can obtain better LDR images with better visual effects,and the colors are more vivid and closer to the real natural world.