The recognition of an edge on a light-and-shadows image captured by a camera is a necessary precondition for all techniques that involve the detection, measurement, or processing of an object. Edge detection technique is therefore of major economic importance. In industrial image processing, an entire block of the above-mentioned techniques is used. Here the central point is to detect whether there is an edge in the test area at all and to localize the edge when it is known to exist. Most edge recognition methods [2, 3], however, presume that an edge does already exist in the test area, and the task is to detect it as precisely as possible. In reality and primarily in defect detection, the potential location of an edge must be determined first. Only then can an edge be successfully scanned for and located. Besides that, real boundary conditions can aggravate the detection of an edge, such as brightness fluctuations of the scanned edge (e.g., different local brightness values at the edge, as with wood), sharpness of the edge representation (e.g., a cant), and the complexity of the edge (e.g., double edge as in a wood board with bark).
KeywordsEdge Detection Grey Scale Test Area Surface Brightness Wooden Board
Unable to display preview. Download preview PDF.