A B C D E F G H I J K L M N O P Q R S T U V W X Y Z All
Content-based image retrieval systems are the challenging area of research in computer science. A retrieval method which uses color feature is implemented in this paper. Effectiveness of various distance measures such as euclidean distance, histogram intersection, histogram quadratic distance and canberra distance are also implemented to determine similarity between feature vectors. Results are compared to determine performance of each of the distance measure during image retrieval. The experimental results show that histogram quadratic distance measure produced more accurate results for image retrieval as compared to other distance measures.
Color is a lower level feature, robust in nature and itsappearance changes as view angle changes. In color image processing each pixel in image is taken into account. In this paper various measures of color distribution implemented. These are color histograms, color coherence vector (CCV), annular, angular and hybrid histograms and are compared in two different color spaces HSV and HVC using average precision recall. Experiments show that CCV and hybrid histograms produced better results in HSV color space.
