Angle Detection And Pattern Recognition Applications

A 2D image feature detection algorithm (KM) and a fast eye location algorithm which is not sensitive to rotation are described in this paper. As we know, an eye location algorithm can be divided into two stages: location of eye window and location of the exact position of eye, and location of corner of eye is critical to the latter, hence KM can be considered .as the preparing stage of the eye location algorithm. PCM, an algorithm based on brightness comparisons and Feature coding and mapping, is introduced first. Its purpose is to locate 2D image feature, including corner and edge. The algorithm combines the concept of binary feature coding and mapping with the concept of brightness comparisoa Because the feature is encoded under a relatively small scale, the algorithm needs oriented template and local feature distinguishment. During the performance test and the comparison with other popular algorithms based on brightness comparison, KM performs well with good accuracy and stability, and it is faster to compute. KM has also advantage of algorithms based on brightness comparison: good location and resistance to noise, so it is more suitable for real time computation.