Choosing the quality of two dimension objects by comparing edge detection methods and error analysis

Abstract

Choosing a quality image is the goal of image processing of two-dimensional (2D) images through computer vision. Image processing consists of stages, namely acquisition, pre-processing (enhancement), segmentation, representation and description, as well as introduction and interpretation. Edge detection is a stage in image processing that aims to find the pattern of an image. This study analyzes the quality of 2D images through edge detection techniques with a comparison of various techniques and error analysis. The comparison of edge detection in this study was performed on images produced using some techniques, such as Canny, Sobel, Prewitt, and Roberts. To analyze the error, Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR) were used. This study was conducted using Matlab by comparing six different images of lung, car, leaf, apple, cat, and motorcycle. The results show that using edge detection with the Canny technique may result in the best MSE and PSNR values. Consistent results of six images detected also show that Canny technique produced the best MSE and PSNR values among the results produced by the Sobel, Prewitt, and Roberts techniques.