Edge Detection in Scanning Electron Microscope (SEM) Images using Various Algorithms

In Image Processing Edge detection is very important. It is an essential technique to improve image quality in various fields including medical, space, biological materials etc. Images are generally represented by its intensities either in grey scale or in color scale. Scanning Electron microscope providing information about images but during deep analysis, users are not able to analyze the problems such as voids, cracks and fiber pullout. Edge is an important part of the image. The extraction of image should not change any features in extracted images. In our proposed work three algorithms are chosen for edge detection i.e. Sobel, Laplacian and Hough Transform detection algorithm is used to detect the edges and removed noise at the same time in SEM images for both metal and non-metal images. Finally, the comparison was done for these algorithms by using various parameters like the speed of execution, noise removal and accuracy in edge detection. In this paper it is concluded, Laplacian gives better result compared to other detection algorithms in terms of voids, boundary, lines or curves and orientation detection.