| It is considered that the target detection is being a very complex question, and is also one of bottleneck of image theory development. The theory proposed on wavelet transform and the fractal has broadened people's horizons. The target detection based on the wavelet transform takes advantage of the multi-resolution characteristic of the wavelet transform to reduce the noise disturbance coming from the image. The target detection based on the fractal theory is that detecting the man-made target detection under the natural background utilizing the discrepancy of fractal characteristics between the man-made target and the natural background.In this dissertation, it has mainly studied the methods of target detection based on wavelet transform and the fractal theory. Firstly the target detection based on the wavelet transform has been studied. It analyzed algorithms of pre-processed of image and improved the target detection method based on the wavelet transform and the energy overlapping to make the result be smaller of the noise disturbance. Then the improvement for the method of computing fractal dimension of images based on the "blankets" and the method of target detection based on fractal dimension have been made after analyzing the fractal theory.However a target detection method based on just wavelet transform or fractal theory can't detect the target accurately enough. A method of target detection based on the wavelet transform and the fractal characteristics has been proposed in this dissertation. This method takes advantage of features of them that is taking the wavelet transform of image, and then obtaining the fractal characteristics from all of bands of the image, and then making the cross-energy operations among all of bands, which can reduce the noise and make the target prominent, at last making image segment through a exact threshold. The experiment proves that the method makes the result of the target detection be more precise. |
PDF Full Text Request