Research for Infrared Image Target Identification and Tracking Technology
|School||Harbin Engineering University|
|Keywords||target recognition image segmentation feature extraction combined moments support vector machine|
Along with the rapid development of the computer technology and digital image processing technology, using the rapid development of computer vision technology for the identification and tracking moving target, has become an advanced and effective technical means. Compared with the traditional manual mode of identification and tracking target, it has good accuracy, intelligence and high efficiency. Because of infrared thermal imaging can work all-day, and the infrared imaging equipment has advantages as strong ability to detect, function of distance and so on. So with the development of infrared radiation theory, infrared detector, infrared optical etc, using infrared technology development of target identification and tracking of methods study more and more attention. It also has become an active field of research.Based on For the purpose on researching infrared image of target recognition and tracking algorithm in complex circumstances, then according to the characteristics of it target image, put forward and design target recognition of infrared image tracking system. Thesis mainly work include infrared image characteristics analysis, image preprocessing, target segmentation algorithm, target recognition tracking algorithm and system software to realize, etc. Concrete from the following several aspects mainly.1. The detailed analysis of the infrared image and histogram characteristics and the image preprocessing algorithm based on image filtering of histogram equalization, and image of sharpening algorithm, the purpose is to strengthen contrast; And the different threshold segmentation algorithm, the comprehensive advantages and disadvantages compared to bring forward a performance good threshold segmentation algorithm to improve visual effect, for the next target recognition tracking preparation. Each of them a algorithm is improved the simulation result, and use of the infrared image processing experiment.2. The BP neural network identification method of neural network, and the structure design. The detailed analysis of the BP neural network algorithm, this paper brings forward the defects of the improved algorithm thoughts. After the division of the binary image information extraction target moment invariant features as a network of input vector invariant moment, and improved. Using the improved invariant moments of the neural network training, the system of recognize performance indexes reaching infrared image of target recognition system requirements. 3. Based on projection method based on adaptive kalman tracking algorithm, by establishing the system model, and in the projection method for target coordinate system the exact position of target, forecast and prior estimate. To achieve the target tracking under the background of night vision.4. Based on the previous study, develop target recognition applications software by VC++language, the software implements the proposed scheme, the results of test show that the software has good reliability and high recognition rate, the operation of the system is stable.