We present a method for detecting and labeling the edge structures in digital gray-scale images in two distinct stages: First, a variant of the cubic facet model is applied to detect the location, orientation and curvature of the putative edge points. Next, a relaxation labeling network is used to reinforce meaningful edge structures and suppress noisy edges. Each node label of this network is a 3D vector parameterizing the orientation and curvature information of the corresponding edge point. A hysteresis step in the relaxation process maximizes connected contours. For certain types of images, prefiltering by adaptive smoothing improves robustness against noise and spatial blurring.

We propose a set of multiscale features for texture description. Image smoothing at multiple scales using the fast smoothing B-spline transform is performed and a number of features, such as the local area, the normal vector dispersion and the gradient orientation, are computed from each scale. Classification experiments with natural textures and comparison with other texture description methods prove the capabilities of the proposed features. Being effective even for small observation windows, these features are suitable for high-resolution texture segmentation.

We present a method for detecting and labeling the edge structures in digital grey-scale images in two distinct stages: 1) a variant of the cubic facet model detects location, orientation and curvature of the putative edge points; and 2) a relaxation labeling network reinforces meaningful edge structures and suppresses noisy edges. Each node label of this network is a 3D vector parametrizing the orientation and curvature of the corresponding edge point. A hysteresis step in the relaxation process maximizes connected contours. For certain images, prefiltering by adaptive smoothing improves robustness against noise and spatial blurring.