Evaluation of Color Spaces for Feature Point Detection in Image Matching Application

In image registration and retrieval developments, feature point detection is essential to find areas in which descriptors are intended. Most of the existing methods use only the intensity information of the images to find the feature points. We inspect the use of color information in feature point detection. Color Information in images is expressed using various color spaces like RGB, HSV, XYZ, LAB, Opponent, YIQ, Y CbCr, and CMY. Deciding the most appropriate color space for a particular application is an open problem in color image processing. To alleviate this problem, a Hybrid color space, with the best features is used for image matching. Feature selection is done using Principal Component Analysis (PCA). Harris corner detection is applied on color images, represented using different color spaces. The feature detection method is compared for viewpoint, rotation, blur and illumination changes. All the experiments use total number of feature points and repeatability measurement for the evaluation.