Computer based target tracking

Today by development of the computer technology and the image processing techniques, target tracking becomes easier in military and civilian area.The most commonly used methods for trackig point detectors (Moravec, Harris, KLT and SIFT), background substraction and segmentation methods. They have some advantages and disadvantages with respect to each other. At this workshop, we choose SIFT method which is the best static/mobile target tracking method in the constant/variable background. SIFT program implemented in MATLAB was used in target tracking. The effects of SIFT parameters in target tracking were analyzed and optimum values of these parameters were obtained. These values are; the number of levels s=3, the necessary threshold value for discarding low contrast points |D(x)|=0.02, the necessary threshold value for discarding edge points R=10, sampling area nxn=4x4 and the number of SIFT descriptors r=8. These optimum values were found compatible with the literature.Moreover, additional program has been written to minimize the probable false matchings between the picture frames. When the program was used without optimum SIFT parameters, it was obtained that the deflection in target tracking was decreased.The target tracking was analyzed by varying picture resolution and the speed of computer processor. When picture resolution was increased; it was seen that the number of SIFT keypoints and the duration of finding SIFT keypoints and the target were increased exponentially. When the speed of computer processor speed was increased, the time required to find SIFT keypoints and the target were decreased.It was concluded that, the static/mobile target tracking has been provded using static/mobile camera with optimum SIFT parameters in the constant/variable background.