Application of Edge Detection Technique for Surface Roughness Estimation of Ti-6Al-4V Turned Surfaces

Abstract

In this research work, a heuristic method based on biologically motivated Particle Swarm Optimization (PSO) has been proposed for edge detection using multiresolution decomposition, to enhance the quality of the images for predicting surface roughness parameter Ra from Ti-6Al-4V turned surface images. First level Dual Tree Complex Wavelet Transform (DTCWT) is used to decompose the turned images to generate new sub band images. The performance of DTCWT with PSO method is examined for turned surface images and compared with conventional edge detectors like Canny, and Sobel methods along with Discrete Wavelet Transform (DWT) with PSO and DTCWT without edge detection. The obtained results showed that, DTCWT with PSO based edge detection provides better looking edges and also best results are obtained in terms of Root Mean Square Error (RMSE) and Peak Signal to Noise Ratio (PSNR). Further, statistical features have been extracted from the images subjected to proposed edge detection method. The extracted statistical features along with machining parameters and tool flank wear have been given as inputs to radial basis function neural network (RBFNN) to predict Ra of the turned surface images.