Fuzzy matching scheme on fourier descriptors for retrieval of 2 dimensional shapes

Shape is one of the most important features in Content Based Image Retrieval (CBIR). When a shape is used as feature, edge detection might be the first step of feature extraction. Invariance to the different transformations like translation, rotation, and scale is required by a good shape representation. In the literature, a large number of shape representations and retrieval methods can be found. In this paper a novel method of shape based retrieval is presented which uses of the modulus and amplitude part of Fourier transform to extract features and to derive the similarity measure. A multi-interval valued type data is used for shape feature representation. Several experiments are conducted on different categories of test images to show the effectiveness of the proposed method. From the results of the experiment, it is clear that the proposed method gives superior precision and recall values compared to a popular work in this field and is invariant to different transformations mentioned above. Also the algorithm proposed in this paper is efficient with respect to computational time.