Subpixel Edge & Line Extraction
MVTec’s software products offer several techniques for sub-pixel precise edge detection and line extraction including:
- Canny, Deriche, Lanser, and Sobel edge detectors
- facet model and Steger line detectors with an accuracy of up to 1/50 pixel
- subpixel color edge and line detectors
- subpixel threshold
Subpixel Contour Processing
Contour processing: Affine and projective transformations; segmentation into and fitting of lines, circles, ellipses, and rectangles; merging of collinear contours; set operations (union, intersection, difference); creation of different standard shapes (circle, ellipse, line, rectangle).
Shape features: Area, center, orientation, circularity, compactness, contour length, convexity, elliptic axis, moments (arbitrary order), eccentricity, hulls (convex, circle, rectangle), and selection of contours based on shape features.
MVTec’s powerful contour processing methods even allow for contours to be defined as Regions of Interest (ROI) to which subsequent image processing operators can be applied. This capability significantly improves efficiency and performance by confining the operations to only those pixels of interest.