Research On Target-detection Algorithms Based On Fractal Features

Target detection is very important for control and guide systems. To trace an object, we should detect it first. In this dissertation, target-detection algorithms based on fractal features are studied grounded on fractal theory.Target-detection methods grounded on differences of fractal features between man-made targets and natural backgrounds have been brought forward, as target-detection method based on fractal dimension, target-detection method based on fractal fitting error, and target-detection method based on D-dimensional area. These methods have simple detection process and easy implementation in comparison with traditional target-detection methods, but still have complicated computation as a result of the complicated computation of fractal features.In this dissertation, a new computation method of fractal dimension has been presented. To compute the fractal dimension, only the amount of the"blankets"needs to be counted. The parameters of the"blankets"are not necessary. The"blankets"here are gray levels cut into different layers under different scales The break between the target parts and the corresponding adjacent parts of the background is used to find out the target. Detection results show that our target-detection method has accurate detection, simple computation, simple detection process, and easy implementation.