Contourlet Transforms For Feature Detection特征检测Contourlet变换.pptVIP

Contourlet Transforms For Feature Detection Literature Review Wei-shi Tsai March 6th, 2008 Feature Detection Focus will be on edge detection Gradient operators (Sobel, Roberts) Laplacian operators LoG (Laplacian of Gaussian) DoG (Difference of Gaussians) Canny method Anisotropic diffusion 2-D Wavelet Transform Separable extension of the 1-D wavelet transform Can be used for edge detection depending on the scale level and the noise in the image Can only detect edges in the horizontal, vertical, and 45° orientations. What features are needed? Multiresolution Localization Critical sampling Directionality Anisotropy 2-D separable transforms satisfy the first three requirements. Curvelets (Candes and Donoho, 1999) Contourlets (Do and Vetterli, 2005) Captures smooth contours and edges at any orientation Filters noise Derived directly from discrete domain instead of extending from continuous domain Can be implemented using filter banks Contourlet filter bank Decomposition The multiscale decomposition is handled by a Laplacian pyramid. The directional decomposition is handled by a directional filter bank. 2-D Discrete Wavelet Transform Contourlet Transform * * First fixed transform to capture contours. However, it is defined in the continuous domain. The transform decouples the multiscale and the directional decompositions.