Object Recognition Using K-Nearest Neighbor Supported By Eigen Value Generated From the Features of an Image
Assistant Professor, Department of Computer Science and Applications, Rathinam College of Arts and Science, Coimbatore, India.
|Related article at Pubmed, Scholar Google|
In this paper, an object recognition system is proposed, that provides the best way to recognize the object from the given image. The process of the proposed method is the input given to the system is the color image. First the color image is converted into Gray-scale image using color conversion method. To obtain the important details of the object, canny’s edge detection method is employed. The edge detected image is inverted. From the Inverted image the feature vector is constructed with the following information i) Hu’s Seven Moment Invariants ii) Center of the object iii) dimension of the object. The information stored in the database for each image is eigenvalue generated from the computed feature vector using Principal Component Analysis. K-Nearest Neighbor is applied to the feature vector to recognize the object by comparing the information already available in the database to the eigenvalue of new image. To prove the efficiency of the proposed method the feature vector generated using orthogonal moment is used with K-Nearest Neighbor Method. The K-Nearest Neighbor outperforms well when compared to Fuzzy KNearest Neighbor and Back Propagation Network.