Research Of Infrared Video Moving Point Target Recognition

It is very hard to identify the weak target from the collected infrared video incomplex background in the real environment. A new small target recognition algorithmbased on energy accumulation and track correlation is used. which achieved good effectto the acquisition of specific infrared video motion point target recognition.Firstly, the algorithm based on suspected target extraction through the analysis ofcomplex background weak point target and background noise, Using software to realizeseveral classic corner detection algorithm and algorithm put forward in this paper,analyze and compare the experimental results some of the infrared sequence imageregistration experiment.Then, the motion point target recognition algorithms based on the direction of thedynamic programming accumulated sequence image. After the suspected targetextraction and rectification, the image includes possible target and related information,first of all, multiple frames make up a two-dimensional projection, then, two timesrecognition with the direction of filter and track correlation accumulated based on thepoint target motion continuity and regularity. When the point target lose, we use the LSto target predictor for track prediction, continue to observe the target, and the forecasttrack for matching and judgment, confidence, and finally recognize the real target, thispaper also to other unusual on the processing.The above algorithms is realized by VC++, and acquisition under the environmentof real infrared video on the above ideas and algorithm testing, the results show that theproposed algorithm for the collected real infrared video object get a great recognizeresult.