Edge detection of infrared targets under complex background based-on fractal theory and dynamic programming

With the view of the difficulty that an infrared target is segmented accurately under complex background. An edge detection method based on dynamic programming (DP) is proposed. Firstly, optimal control function based on different fractal dimensions and gray level features is created. The second step is to seek the latent object point by locally adaptive threshold technique and to calculate the minimum cumulative cost matrix. Finally the optimal and connected contours are determined by DP theory. Experiment result with real targets show that the method can detect the target effectively and reduce the false alarm probability.