Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

A Study on the Technology of the Infrared Image Matching

Author WangSiZuo
Tutor YangChunLing
School Harbin Institute of Technology
Course Electrical Engineering
Keywords Image matching Cross-Correlation NMI characteristics DSP
CLC TP391.41
Type Master's thesis
Year 2008
Downloads 242
Quotes 0
Download Dissertation

In modern high-tech war, the main form of the war is the long-range strike of precision guided weapons to achieve the objective of the war with a minimum price for the greatest achievements of war. One of the important forms of precision guidance technology is infrared imaging guidance; also one of the key technologies of it is infrared image matching technology, which has always been a hot and difficult topic for people. The aim of image matching is to search for one or more transformation in the transformed space, and to align two or multiple images of the same scene under different times, different sensors and different perspectives. Recently due to the extensive application of cruise missiles, foreign countries have adopted scene matching technology in order to enhance cruise missiles terminal guidance, so fast image matching technology is very worthy of study.This paper analyzes the matching technologies of the infrared image matching and propose the improvement schemes.Firstly, the paper puts forward a fast-matching method with the pyramid hierarchical search idea in order to improve the real-time processing and reduce calculation. As different similarity measurement methods are applied to different resolution images. Compared to the means removal relevant matching algorithm performance, this method has a good performance on anti-jamming and anti-distortion under variable illumination and mild rotation.Secondly, a pixel-based counterpart’s similar measure matching algorithm is put forward to solve the image problem of noise and partial block. And the paper analyses the matching tracking performance of this algorithm compared to the minimum error measure (AD), respectively in the absence of noise, noise and block the three local cases.Then, the paper studies the application of normalized moment of inertia of the image matching for a certain image and gray geometric distortion of the situation. The experiments certificate its anti-gray distortion and better invariance in the geometric distortion (such as translation, rotation, zoom), and high retrieval correct rate, has smaller amount of computing and higher speed compared to the same moment. Finally, the paper combines this issue in research and DSP hardware processing platform effectively, and makes full use of DSP chips work performance, realizing the rapid image match taking the advantages of hardware.

Related Dissertations
More Dissertations