Video Gun Forensics Algorithm And Medical Imaging Platform Interaction Design

With the development of technology in image processing and computer vision, vehicle speed estimation (VSE) through video signal is used in ITS (Intelligent Transport System) more frequently.This paper focuses on the VSE algorithm design on an embedded video processing system equipped with two cameras. The VSE problem is concerned in two steps. Firstly, we distill features simultaneously from synchronous images generated by the dual-camera system. Secondly, feature matching with high accuracy is required between the parallel features. In this paper, an accurate and fast method to match the license plate in images of the vehicle is proposed.Chapter 1 introduces the advantages and disadvantages of the methods to estimate the speed of vehicles that used widely nowadays, and gives the basic concepts of image processing and computer vision based on which we develop our methods.Chapter 2 is about the hardware and software construction and execution flow of the system.In Chapter 3, a robust and efficient background abstraction algorithm based on MRF is proposed.Chapter 4 introduces a method we use to calibrate our camera, one which is based on the famous Two Step Method.Chapter 5 firstly presents the method we allocate the license plate in the images and then introduces how an accurate parallax is computed from them.In Chapter 6, we discuss what Kalman Filter can do in terms of increasing the system accuracy.Chapter 7 begins a new topic about the design of interaction functions in IMIP (Integrate-platform for Medical Image Processing).Chapter 8 concludes the paper.