Rotating kernel transformation based edge detection using adaptive threshold
Vasagiri Venkata Guruteja, Mantosh Biswas · 2016
Edge detection is the first step in many computer vision applications. Edge detection significantly reduces the amount of data and filters out unwanted or insignificant information and gives the significant structural information of an image. We proposed an edge detection algorithm using weighted kernels with adaptive threshold values. Instead of using a constant threshold value we use dynamic threshold value based on local image information and the kernel used. It was observed from the experimental results that our proposed algorithm produces better results for the real time images having non uniform intensity regions minimizing the false positive and false negative edges.