De-noising And Super-resolution Reconstruction Algorithm Research On Range Image

PMD camera based on TOF (time-of-flight) principle can capture depth image of the scene in real time, and has broad applications in three-dimensional visualization, face detection, video production, vehicle safety and other areas. For the limitations of the PMD camera, this thesis focuses on de-noising and super-resolution reconstruction algorithm for PMD range image. The research of this thesis has an important practical value.In this thesis the main work is as follows:(1)Based on the working principle of PMD camera, this thesis analyzes the cause of the error and the de-noising methods for range image.(2)Based on "adaptive normalized convolution", we realize the improved "adaptive bilateral normalization convolution" method by using amplitude image as confidence and increasing the application of the amplitude information in range image de-noising. Experimental results show that the improved algorithm has a well edge-preserving, and can remove the noise effectively.(3)On the basis of BTV (Bilateral Total variation) super-resolution algorithm, combining the characteristics of motion matrix and blur matrix, we realize the fast BTV super-resolution reconstruction by finding a blurred HR image from the LR measurements and estimating the de-blurred image from the blurred HR image. Experimental results show that the improved algorithm not only improves the quality of the range image, but also reduces the running time of the algorithm.