Reliability Performance Comparison of Different Wavelet Transform Based Facial Image Retrieval Under Partially Occluded Faces

摘要

In this paper, the reliability performance comparison of Different wavelet transform features such as Discrete Wavelet Transform with PCA, Lifting Wavelet Transform with PCA, Dual tree complex Wavelet transform with PCA and Gabor wavelet Transform with HMM on partially occluded face images such as Eye brows, Eyes, Nose, mouth, chin occluded test face images are performed. These approaches are tested on ORL database at different levels with Euclidean distance classifier. The reliability performance for Gabor wavelet Transform with HMM based method was found equivalently efficient in many cases. The better performances of the proposed algorithms are based on Average retrieval accuracy, weighted retrieval accuracy and Standard Deviation Values.