Efficient Shape Classification using Zernike Moments and Geometrical Features on MPEG-7 Dataset

摘要

There is an urgent need and demand for manipulating images to extract useful information from them. In every field, whether it is biotechnology, botany, medical, robotics or machinery, the demand for extracting useful aspects of a specific targeted image is growing. Effective systems and applications have been introduced for this purpose: CBIR and MPEG-7 are most common applications. Shape extraction and recognition is used in image retrieval and matching. Complex objects can be identified and classified by extracting their shape. This paper proposes an efficient algorithm for shape classification. Analyses are made on MPEG-7 dataset using 1400 images belonging to 70 classes. Zernike Moments descriptor and geometrical features are used for classification purposes. Cross validation and percentage split are used to evaluate the proposed scheme. Experimental results proved the efficiency of the proposed approach with an accuracy of 92.45 percent on the challenging dataset.