The Use of HSV Color Model for Subtle Ear Region Extraction


Identifying people by their ear has recently received import attention in the literature. The accurate segmentation of the ear region is vital in order to make successful person identification decisions. This paper presents an effective approach for ear region segmentation from color ear images. Firstly, the RGB color model was converted to the HSV color model. Secondly, thresholding was utilized to segment the ear region. Finally, the morphological operations were applied to remove small islands and fill the gaps. The proposed method was tested on a database which consisted of 105 ear images taken from the right sides of 105 subjects. The experimental results of the proposed approach on a variety of ear images revealed that this approach is efficient in segmenting ear images under various environments and it gave a detection rate of 98.61% for ear segmentation.