Segmentation for Main Body of Typhoon from Satellite Cloud Image by Genetic Algorithm in Contourlet Domain

Abstract

A single-threshold algorithm to segment typhoon cloud image is proposed based on contourlet transform and Bezier histogram. In order to enhance the contrast and reduce noise of the original typhoon cloud image, discrete contourlet transform is done to the original typhoon cloud image. The contrast is enhanced in the low frequency components based on non-linear transform in the contourlet domain. GA is used to obtain the optimal non-linear gray level transform parameters. Noise is reduced in the high frequency components in the contourlet domain. Then the processed typhoon cloud image is used to segment the main body of the typhoon from the typhoon cloud image. First, the gray-level histogram of original image is obtained. In order to reduce noise, the gray-level histogram is smoothed by Bezier curve and Bezier histogram is obtained. A proper threshold is determined in the Bezier histogram to segment the typhoon cloud image. Experimental results show that the proposed method can effectively segment the main body of the typhoon from the typhoon cloud image.