NEURAL NETWORK CORRELATION BASED SIMILARITY EVALUATION WITH ZERNIKE MOMENTS FOR THE POSE-INVARIANT FACE RECOGNITION

Human face recognition is best application in pattern recognition for identification and recognition. Development of face recognition system is increasing day by day in market and research organizations. Different parameters and methods are used for face recognition. In this research project, we will discuss about the different algorithms used for face recognition that are Zernike Moments (ZMs) and correlation classification (CC) etc and compare these algorithms with proposed algorithm Z_CC (Zernike with Correlation Classification).The angular information or rotation of the face is calculated by using the Zernike moments (ZM) to obtain the degree or radian of face rotation from the frontal view. The robust combination of angle-invariant and scale-invariant features with the combination of Zernike moments and correlation classification has been proposed with the neural network classification. The experiments will be performed on the variety of datasets. The multi-object dataset has been combined by collection the samples with faces rotated in the training samples. Z_NN (Zernike with neural network) algorithm provide best recognition rate for human face recognition 90%. In this algorithm we use Zernike Moments and correlation for global feature extraction and after that these features are compared by using neural network.