Study On Infrared Small Target Image Process Technology Based On DM642

Infrared small target usually is a far-distance detecting image of an object in sky. Due to some questions such as few pixels amount, scarcity of sharp and grain information, dim signal intensity, how to recognize and track has become a significant and challengeable research topic in the realm of image process.This project is a research of detecting and tracking infrared small target using Digital Signal Processor. The main goal is to apply high-rate processor TMS320DM642 fitting in with digital media and discuss the infrared small target image process technology.Firstly, refer to actual captured infrared video image this paper processes theoretic and practical test. Furthermore, bring forward arithmetic about single target image in sea-sky background to fulfill detection and tracking. The arithmetic include some aspects as follows: background de-noising based on wavelet and template filter; threshold of binary image transformation using histogram; morphological operation; sea-sky line separation applying on Hough transform and searching oddity point of gray-level only in specific region confirmed by known region information. And the tracking method is Range Gate. These above-mentioned factors compose the whole structure that can finish exact detection of small target.Secondly, this paper completes a system in DM642 hardware test platform. The high-speed real time system of multi-task is based on DSP/BIOS while this project has finished programme referring to Referance Frame 5. The project includes four parts, that is, video data capture; pre-process; detecting and tracking arithmetic and result display. According to Algorithm Standard of Ti Corporation, these program is encapsulated into cells. Moreover, aiming at CACHE apply to optimizing configuration and compile the bootloader program to self-boot. At last, this paper finishes emulation experiment to validate afore-metioned arithmetic that can detect and track infrared small target. Considering to software embedding course, this paper completes performance testing and research that includes main parts of this system. Also the further technology and measure to improve running efficiency are proposed.