A Comparative Analysis of Edge Detection Using Soft Computing Techniques

Detecting edges is one of the most significant aspects of computer vision. Typical methods for edge detection like Sobel and Canny are robust and fast, but they are sensitive to noise. Soft computing techniques such as particle swarm optimization (PSO), ant colony optimization (ACO), genetic algorithms (GA) and fuzzy logic system (FLS) have extensive application in edge detection of images because of their adaptive behavior. Edge detection is identifying the discontinuities in intensity of the pixel and grouping the contour of edges. The quality of edges in ACO-based edge detection majorly depends on the choice of constants, pheromone evaporation rate, number of iterations etc. In PSO-based edge detection, the quality of images depends on the values of acceleration coefficients and inertia weight. However, thresholding is major stakeholder in determining the fitness of the chromosomes. The population contains 2-D chromosomes. Fuzzy systems are most suitable for designing edge detection hardware. This paper presents a thorough comparative study of soft-computing-based edge detection techniques and highlights their key features. The factors affecting quality of edges are compared, and the actual outcomes of the approaches are systematically arranged for better understanding.