Image Tracking Method And Dsp Implementation

Millimeter wave imaging terminal guidance has become the precision-guided research focus, for its all-weather, all day and anti-interference ability, etc. While the image matching using millimeter wave SAR remote sensing images is not only a millimeter wave imaging seeker's key technology, but also the image tracking's key technology. In a variety of target tracking technology, the image tracking technology, with its unique superiority, has been paid widespread attention and been used in many fields. It played an important role in national defense and civils.To solve this problem, image tracking algorithm was studied deeply. On the basis above, for different types of target including buildings on the ground and ships in the sea and so on, an image tracking system based on TMS320C6416 was developed to realize the entire process of matching, detecting and tracking. Specific works are as follows:1. Studied the image matching algorithms deeply. The normalized correlation coefficient algorithm (NPROD) of the image matching was improved to make it more suitable for DSP processing; A new extraction method of Gabor transform image invariant features was proposed for image matching. The matching accuracy was improved on the condition of image distortion .2. Studied the search strategy of the image matching deeply. Genetic algorithm was adopted for the image matching as a search strategy using it's non-traversing property, which greatly reduced the matching time. The pyramid hierarchical search and level-matching method was adopted to solve the real-time problem without impacting the matching accuracy.3. Studied the image matching performance assessment method. A new calculation method of matching confidence level was given on the basis of NPROD, which calculates the ratio of peak and sub-peak of the correlation surface's second-order difference. The experimental results show that the confidence was greatly improved.4. Studied detection and tracking methods for two types of targets including buildings on the ground and ships in the sea. A new method, which calculates the weighted summation of potential targets'indicators, was proposed to determine the correct target. The experimental results show that this method is feasible and effective.5. Developed a real-time image tracking system based on TMS320C6416. System software design was introduced deeply, including image matching module, target detection module and the image tracking module. It made an in-depth study on the system involved in code optimization, memory allocation, large amounts of data to read. Real-time property of the whole system achieved engineering requirements.