High speed edge detection implementation using compressor cells over rsda

Recently computer vision is playing an important role in many essential applications, such as medical image analysis, visual surveillance, etc. Many of these applications are subject to a "real-Time constraint", therefore it requires a fast and reliable computation system. Edge detection is the approach used most frequently for segmenting images based on changes in intensity, it extracts important structural information needed for high-level functionality and reduces the amount of data that needs to be processed. There are various kernels employed to achieve edge detection, such as Sobel, Robert, and Prewitt, upon which, the most commonly used is Sobel. This paper introduces a novel type of operator cells on the Reconfigurable Static Data-flow Architecture (RSDA), which is a scalable architecture optimized for the computation of image and video. This enhancement shows significant improvement, as it decreases the computational 26%, compared to using the conventional adder cells, and also decreases the LUTs and hardware resources of the architecture. A comparison between the conventional adders and different types of compressors has been exploited based on results from simulation on Isim simulator and a flooring plan using PlanAhead tool.