Comparison of data fusion in different color spaces for color iris recognition

In this paper, we first introduce some fusion strategies to color images based iris recognition. They are used to combine the matching scores of different channels. Their performances are compared in two color spaces: RGB and YCbCr, together with another artificial space, namely RYCr. The experiments show that in many cases, multiple channels fusion can help to improve the performance compared to the best single channel. The Sum, Product and Weighted Sum rules are more robust than the Min and Max rules. The effectiveness of channels fusion is more remarkable in YCbCr than that in RGB and RYCr, which confirms that independent information may be more helpful than dependent ones, even when some of them are poor.