MIRS: Museum Image Retrieval System Using Most Appropriate Low-Level Feature Descriptors
The main component of Content Based Image Retrieval is a feature extraction, where automatically extracted a low-level features (color, texture and shape). In this paper, several features extraction methods are explored to examine their effectiveness in retrieving images including Color Histogram, Color Layout Descriptor and Color Moment descriptors to represent the color feature. Texture is represented by Gray Level Co-occurrence Matrix, Local Binary Pattern and Gabor filter descriptors. Shape is represented by Hus seven invariant moments and canny edge detection descriptors. A new approach to select the most appropriate descriptors to represent the image as uniquely and accurately using the average of success method is presented. Six transformations is applied to 100 original images of Iraqi National Museum of Modern Art collection to demonstrate experimentally the efficacy of the proposed approach and promising results are reported.
Keywords: CBIR, Feature extraction, Low-Level features, Feature selection, Average of success method
Download Full-Text
ABOUT THE AUTHORS
Fatin Abbas Mahdi
Lecture at Foundation of Technical Education, Iraq and Ph.D. student in Software Department, college of Information Technology, University of Babylon-Iraq
Abdulkareem Ibadi
Software engineering Department, Baghdad college for economic studies university Baghdad, Iraq
Fatin Abbas Mahdi
Lecture at Foundation of Technical Education, Iraq and Ph.D. student in Software Department, college of Information Technology, University of Babylon-Iraq
Abdulkareem Ibadi
Software engineering Department, Baghdad college for economic studies university Baghdad, Iraq