Analysis of Segmentation and Identification of Square-Hexa-Round-Holed Nuts Using Sobel and Canny Edge Detector

Abstract

In the existing real-time automobile shop, it is difficult to trace an object and identify its presence. The failure may happen due to its absence or improper match of shape as its identity. So to overcome this, we propose a method which can be used for the automatic identification of vehicular nuts based on the input image that contains a nut with square, hexa, rounded-head and pinned-bolt shapes. The application even works with the nut having clear view or any entity added like mud, noise, colour, etc. on the surface of nut. For the identification process the database has been designed to store different shapes for selected number of nut-shapes. By applying median filter during pre-processing stage, the Sobel-edge-detector and Canny-edge-detector; segmented and identified the captured images by identifying the edges to ascertain shape of the input. With the experimentations carried the method results with an accuracy of 86.1875%