A comparative study among colorful image descriptors for content based image retrieval
Kamlesh Kumar, Jianping Li, Zain-Ul-Abidin, Imran Khan · International Conference on Computing for Sustainable Global Development · 2016
Recently there has been an explosive growth of image archive libraries on web media. Due to this fact image retrieval technology has achieved significant outbreak and many Content based Image Retrieval (CBIR) methods have been proposed during last few years. CBIR is a kind of image mining technique which analyses the image features using color, texture, shape and object position. In this paper, research has been carried out on color feature extraction. For this purpose, a comparative study among three colorful image descriptors namely Columnar Mean, Average RGB and Color Moment has been discussed. A CBIR system has been designed for each method separately and also their retrieval performance have been evaluated through computation of average precision, average recall and f-measure. Euclidean distance has been employed for matching reference image with database images. The simulated results showed that average RGB color image descriptor outperformed than other two color feature extraction methods.