Comparison And Analysis Of A Typical Atr Experiment Algorithm

Automatic Target Recognition (Abbreviated ATR) is an important research topic in the field of intelligent machine vision. This paper takes the research ideas of automatic target recognition, and uses sub-modules ideas researching, analysing and experimenting typical ATR algorithm. Under certain conditions, this paper solves the problems of data acquisition for specific image, data storage, target detection, target recognition and performance contrast while meeting the real-time property.The contents of this paper include:the image acquisition, target detection, feature point extraction, object recognition and algorithm evaluation. This paper uses median filter to remove image noise points of pre-processing and post-processing in image preprocessing, and then it uses the contrast algorithm to improve the contrast difference of the target and background. Target detection uses a variety of segmentation algorithm to extract the multi-target data, and it uses centroid to position regions after extraction. Corner detection algorithm takes comer detection for image, through probability density of the corner points to improve the time issue of overall comers calculation. This paper uses two algorithms for target identification:1) Improved gray value template matching algorithm;2) Feature points matching rate algorithm; where, improved gray value template matching algorithm templates include standard data already existing in source database and template data with the training renew process. Extracting feature points mainly uses feature corner points factor to obtain HARRIS data, by calculating the matching rate of matching comers and setting threshold for matching rate to identify the target.By the OTC texts, this paper achieves the experimental comparison and analysis for typical ATR data. This paper integrates a variety of modules processing algorithms and2D curve tracing functions. It provides the basic idea reference for large ATR data experimental platform, it also has some practical value.