A novel face recognition method based on the local color vector binary patterns of features localization

Abstract

LCVBP (Local Color Vector Binary Patterns) approach extracts multi-signal channel characteristics from color norm patterns and color angular patterns of a color image. As a result, feature dimension is higher and computational cost is greater. Hence, this paper presents a novel region-based LCVBP feature extraction method for face recognition. Firstly, we locate the feature points in a face image, such as eyes, nose and mouth, and obtain feature region by utilizing the location of feature points. Secondly, the LCVBP histograms of these feature regions are extracted, and sequentially put together as the final histogram characteristics of an image. Experimental results show that by abandoning this redundant information in a face image, we can also obtain the approximately equal identification rate with the LCVBP approach, but the dimension of characteristic vector is reduced greatly, the calculation cost is reduced significantly, and face recognition can be achieved faster.