Edge detection using multiring on spiral architecture

Improving the computation efficiency is the key issue in image processing, especially in edge detection, because edge detection is very computationally intensive. With the development of real-time image processing application, fast processing response is becoming the major requirement in this area. In this paper, a parallel and distributed algorithm on Spiral Architecture for edge detection using a particular network, namely MultiRing is proposed. The proposed algorithm is based on Master-Slave model. However, it guarantees a better load balancing among the processors (or nodes) and greatly reduces the traffic from and to the master node. It also requires less processes be created on the master node and hence saves CPU time for less process switches. Our work sees a potential for further speeding up distributed image processing such as object recognition and pattern recognition based on computer clusters.