Research Of Target Detection And Recognition Of Infrared Images Based On The Active Contour Model

In recent years,with the rapid development of IR thermography system,the infrared image target detection and recognition technology has been the important research topic in the field of modern image processing.This technology was played a significant role in many fields such as military,medicine,monitoring and transportation.This paper was analyzed the core problem of the infrared image target detection and recognition technology and conducted the thorough research of this technology.Firstly,the commonly used detection algorithms were studied in this paper.The edge detection algorithm,the growth algorithm based on area and the active contour model were mainly studied,and a lot of simulation and realization experiments was conducted for these algorithms.Secondly,based on the combination of adaptive edge detection and active contour models,an improved algorithm for extracting target contour of infrared images automatically was proposed to deal with troubles which were being more sensitive to the position of initial contour and not converging to concavity target contour correctly.The emulation results and analysis demonstrated that,when using this method,the infrared target contour could be converged autonomously,the ability that the initial contour was converged to the real boundary accurately could be enhance,and the accuracy of detection was improved.For the infrared image target recognition technology,the feature extraction methods such as the geometric invariant moments and Hu invariant moments were mainly studied firstly,and a lot of in-depth analysis and research was conducted for verifying its stability and distinction.Secondly,the minimum distance classifier and the Bayes classifier were mainly studied.Then the feature quantities such as the geometric invariant moments and Hu invariant moments and classifiers such as the minimum distance classifier and the Bayes classifier were simulated and analyzed.Finally,the infrared targets recognition algorithm based on the Bayesian Network was proposed.The experimental results show that the recognition rate of the propose algorithm was the highest,and this algorithm has a better recognition effect.Finally,based on the software platform which was constructed by VS2010 and OpenCV2.4.3,the algorithm for extracting target contour of infrared images automatically was realized,and the feasibility of this algorithm were verified by experiments.The experimental results show that the infrared target contour could be converged accurately and autonomously when using the algorithm for extracting target contour of infrared images automatically which was proposed.