Object contour detecting using pseudo zernike moment and multi-layer perceptron

Abstract

Most of the object contour detection approaches suffers from some drawbacks such as noise, occlusion of objects, shifting, scaling and rotation of objects in image which they are failed to recognize object contour. A solution to solve the problem is utilization of feature extractor which is independent of rotation, scaling, transformation and etc. One of the best options for the task of feature extraction is Pseudo Zernike Moment (PZM) which is able to extract feature vector in all situations. In this paper, a new method of facial age estimation is proposed which includes three steps: object region detection, feature extraction and contour detection. The first step employs Line Detection with Contours (LDC) in order to find the object region based on the connected components inside the image. In the second step, PZM is applied on the previously detected object regions to extract feature vector. At the final stage, Multi-Layer Perceptron (MLP) model is employed to extract final object contour based on the PZM-based feature vector. Experimental results on Caltech-101 dataset shows that classification rate is improved to 98.47 which proves the ability of the proposed method.