Infrared small target detection by improved fractal and local entropy

This paper describes a new fractal algorithm,integrating three fractal characteristics,fractal dimension,intercept and fitting error to handle fractal characteristics of each pixel. The algorithm reduces defect of the traditional fractal method in distinguishing between target and background. On account of that fractal method is a heavy computation,we exploit local entropy theory to obtain a small area of the target,detecting by fractal dimension analysis. Experiment results prove that the proposed method can effectively reduce false alarms and improve the detection rate.