FPGA-Based Sobel Edge Detector Implementation for Real-Time Applications

Abstract

This paper describes an implementation of real-time edge detection algorithm based on Field Programmable Gate Arrays (FPGA). The presented block is the first step for an embedded application for automotive system. From the nature of the application, challenges were hard real time constraints, robust edges detection with optimized hardware resources. In order to respect the requirements our choice is an implementation of the Sobel detector for its better compromise noise/material resources. Furthermore, to enhance noise performances of our architecture, we implement a dual threshold instead of a one threshold in the final stage of the detector. The obtained results show that proposed edge detector can reliably detect edges even if in the presence of luminance variation, using optimal hardware resources.