Shape Detection using Geometrical Features

In this paper, we have presented an approach for object detection system. This approach is used for detect two-dimensional shapes such as lines, rectangle, square, circle, triangle, polygon, star etc. Proposed technique of shape detection is based on the statistical properties of distribution of points on bitmap image and sub-windowimage of a shape. For recognition, we have considered sub-windowbased features and Nearest Neighbours classifier. By applying these features, we achieve maximum recognition accuracy of 96.7% using 4556 samples.