PRECISION TRACKING WITH SEGMENTATION FOR IMAGING SENSORS

We present a method for precision target tracking based on data obtained from imaging sensors when the target is not fully visible during tracking. The image is divided into several layers of gray level intensities and thresholded. A binary image is obtained and grouped into clusters using image segmentation techniques. The association of the various clusters to the track to be estimated relies on both the motion and pattern recognition characteristics of the target Using the centroid measurements of the clusters, the probabilistic data association filter (PDAF) is employed for state estimation. Expressions for the single-frame-based centroid measurement noise variance of the target cluster, and the optimal parameters for cluster segmentation are given. The simulation results presented validate both the expressions for the measurement noise variance as well as the performance predictions of the proposed tracking method. The method is first illustrated on a dim synthetic target occupying about 80 pixels within a 64 x 64 frame in the presence of noise background which can be stronger than the target. The target is modeled as having an intensity distribution within a narrow range and the background in a much wider range, both above and below the average intensity of the target. The binary image obtained after an ''intensity bandpass'' thresholding is reduced to clusters by the nearest neighbor criterion. The results show that it is possible to achieve subpixel accuracy in the range of 0.3 to 0.4 pixel rms error with moderate (0.7) to low (0.3) target pixel detection probability. The subpixel accuracy can be further improved for a larger target The usefulness of the method for practical applications is demonstrated by considering a sequence of real target images (a moving car) where the measurement noise variance was calculated as having 0.7 pixel rms value. The achieved filter accuracy for position was 0.4 pixel rms in each coordinate and 0.09 pixel/frame for velocity after 10 frames.