A local edge detector used for finding corners

Many corner detectors make use of edge detectors to extract edge points first. In this paper, we proposed a local edge detection algorithm that can be used well in corner detectors. Instead of marking edge points globally, we detect them locally based on both gradient magnitude threshold and gray level analysis. The process is dynamic and can provide correct edge points for detecting corners. The problem of finding corners is simplified into detecting simple lines (straight lines that pass through the coordinate origin) in a local coordinate system. Hough transform is modified to organize the edge points generated by the new edge detector into simple lines. Experiments proved that our new corner detector based on the local edge detector works well over most images, and is fast enough for real time applications.