Research On Image Processing And Tracking Algorithm Based On PICO384

Infrared imaging guidance technology has been widely applied in areas such as precision guidance and navigation for its advantages of high precision,all-time,all-weather,good concealing and strong anti-interference.Infrared target tracking system is one of the key technologies of the infrared imaging seeker.Around IFPA(Infrared Focal Plane Array)PIC0384,we realize the real-time infrared image preprocessing algorithm based on FPGA in this paper in order to extract target feature further in the follow-up tracking system,and a single infrared target tracking system is developed as well on that basis.This paper introduces the general development situation of un-cooled infrared imaging technology and infrared target tracking technology firstly,as well as the prospect of their applications.Then we expound the principle of un-coo led infrared detector,the characteristics of infrared image,image processing algorithms and the infrared target tracking technology.On the basis of these theories,we construct the infrared image processing and target tracking hardware platform based on the PIC0384 IRFPA.On the basis of the existing hardware of 384×288infrared imaging system electronic components,we realize the real-time infrared video processing algorithms regarding combined ways in detecting and compensating blind pixels and Successive Mean Quantization Transform(SMQT).Compared with histogram equilibrium algorithm,the SMQT enhancement algorithm can enhance the image contract,meanwhile preserve more image details,and effectively prevent problems occurring in the traditional enhancement algorithm such as over enhancement and "patches".After getting the clear real-time infrared image,we add a single target tracking system based on FPGA and transfer IR video stream to the target tracking system from the imaging system through the LVDS technology.Simulating and verifying the TLD tracing algorithm based on the Shi-Thomasi features by experiment,as well as the tracing algorithm based on SIFT features,we propose a optical flow tracing algorithm based on the improved SIFT features and realize it by hardware.Detecting feature points by DoG(Difference of Gaussian)operator,removing the edge response of SIFT feature points through the local maximum inhibition method,matching by using the simplified SIFT descriptor,obtaining the target location and scale through optical flow algorithm,and stacking the tracking window,we finally realize the accurate real-time tacking and present the hardware implementation steps as well as its related treatment effect.