Fuzzy Heuristics in Content Based Image Retrieval

Abstract

The field of content based image retrieval has gained steady momentum in recent years due to the growth of diverse digital images in the society. This paper proposes a novel method for retrieving images based on color, texture and shape features controlled by fuzzy heuristics. The color feature is extracted using color auto correlogram. Wavelets like Haar and Daubechies are applied for extracting texture features. The proposed method of obtaining histogram value after Canny edge detection is employed to extract the shape features. The feature values are concatenated. The similarity is analysed using Mahanabolis distance. To obtain more relevant results fuzzy heuristics is applied which acts as a controller in presenting more relevant and similar images as retrieved results. The evaluation of the proposed approach is carried out using the precision and recall measures.