Extraction of ocean eddies based on contourlet analysis and morphology

In this study, a combined algorithm using the Contourlet transform and morphology is proposed to detect the significant spatial patterns of ocean eddies. Contourlet transform, which is introduced recent years, has better performs in representation of geometry lines or curves than the wavelet transform. The source image was decomposed by Contourlet with several levles, the high-frequency images is divide into eight directions by the directional filter in each level. Then deal with the high-frequency and low-frequency coefficients separatically, use mathematical morphological method extracts edges in low-frequency approximate. At last, the two edge images were rebuilt to obtain an integrated and clear edge image. The Hough transform is than used to extract the characteristic of the eddies. The experimental results show that this algorithm cabin the priority of the Contourlet and morphological method and is superior to other traditional edge detection method such as the grads method, the Sobel method, or morphological method alone. The study verified that the algorithm proposed is an effective way to identify and detect ocean eddies with complex form.