Depth Image Super-resolution Reconstruction Algorithm Based On 3D TOF Technology

The 3D TOF(Time of Flight)technology has changed the machine vision industry by using low-cost CMOS pixel area arrays and active modulation light source technology to provide distance depth of field information for 3D scenes.This paper describes the working principle of the TOF camera,and analyzes the traditional image filter.Through the comparison of 3D vision technology,it further explains the application prospect of the TOF technology.Combining with the research status of 3D TOF cameras,the 3D TOF working principle and the current mainstream software to optimize the image quality and improve the resolution of the algorithm are studied.The main work is summarized as follows:(1)The depth image degradation model is analyzed,and the depth image super-resolution algorithm is analyzed from both the depth image itself and the color image combined with the same scene.For the method of combining color images in the same scene,according to the prior knowledge of the color image,through the feature information fusion of the RGB-Depth system,mining the intrinsic relationship between the two and guiding the reconstruction of the depth image is one of the current mainstream research directions.The working principle of joint bilateral filters and guided filters is analyzed in detail.(2)Combining high-resolution color images of the same scene,this paper proposes a super-resolution reconstruction algorithm based on depth image confidence and non-local self-similar filtering.Based on the joint bilateral filters,the depth image confidence information is used to denoise the smooth image,and in the non-local self-similar filter,the feature structure information of the same scene color image is used for constraint and global optimization,and the ringing can be well reduced.Effects,reconstruct high resolution images.(3)Combine the high-resolution color images of the same scene in the local structural features and optimization theory,this paper further proposes an optimization algorithm to reconstruct the depth image through the optimal solution of the model.Combining with the optimization theory to establish the energy function,this paper constructs the energy function data item based on the reconstruction constraint conditions,and uses the same target scene depth image and the corresponding color image to have linear similarity in the local domain and construct the smooth function term of the function.The optimization algorithm has a good effect,and has stronger robustness than the traditional depth image filtering algorithm.