Comparative Study on Content-Based Image Retrieval (CBIR)

The process of retrieving desired images from a large collection is widely used in applications of computer vision. In order to improve the retrieval performance an efficient and accurate system is required. Retrieving images based on the content i.e. color, texture, shape etc is called content based image retrieval (CBIR). The content is actually the feature of an image and is extracted through a meaningful way to construct a feature vector. Images having the least distance between their feature vectors are most similar. This paper gives comparison of three different approaches of CBIR based on image features and similarity measures taken for finding the similarity between two images. Results have shown that selecting an important image feature and calculating that through a meaningful way is of great importance in image retrieval. All the important features must be considered while constructing a feature vector and a proper similarity measure should be used for calculating the distance between two feature vectors. These parameters play very crucial role in deciding the overall performance of the any CBIR system. Some future direction were identified and under our future work.