Interactive Image Retrieval with Wavelet Features

Abstract

Abstract. This paper presents an iterative Content Based Image Retrival(CBIR) system with Relevance Feedback (RF), in which M-band wavelet features are used as representation of images. The pixels are clustered using Fuzzy C-Means (FCM) clustering algorithm to obtain an image signature and Earth Mover's Distance (EMD) is used as a distance measure. Fuzzy entropy based feature evaluation mechanism is used for automatic computation of revised feature importance and similarity distance at the end of each iteration. The performance of the algorithm is tested on standard large multi-class image databases and compared with MPEG-7 visual features.

Novel CBIR system based on Ripplet Transform using interactive Neuro-Fuzzy technique

“…The user marks the images returned by the search engine as relevant or irrelevant samples. A fuzzy based relevance feedback algorithm use this feedback information to select a set of better 20 images from the partitioned DB in the next iteration [10]. This retrieval process finishes at a point when the user is satisfied with the retrieved result.…”

“…Most of the RFMs, employ two approaches namely, query vector moving technique and feature re-weighting technique to improve the retrieval results [9]. Feature re-weighting technique uses both the relevant and the irrelevant information, to obtain more effective results [10,11]. But in all these cases, time complexity per iteration are high and accuracy of the relevant images are low.…”

Interactive Content Based Image Retrieval Using Ripplet Transform and Fuzzy Relevance Feedback

“…High retrieval efficiency and less computational complexity, are the desired characteristics of an efficient CBIR system. Wavelet transform (WT) based low level features, provide a unique representation of an image, and are highly suitable for characterizing textures of the image [2]. Many WT based CBIR systems have been proposed in the literature [3,2].…”

“…Wavelet transform (WT) based low level features, provide a unique representation of an image, and are highly suitable for characterizing textures of the image [2]. Many WT based CBIR systems have been proposed in the literature [3,2]. But WT is inherently non-supportive to directionality and anisotropy.…”

“…Most of the RFM, employ two approaches namely, query vector moving technique and feature re-weighting technique to improve the retrieval results [7]. Feature re-weighting technique utilizes both the relevant and the irrelevant result's information, to obtain more effective result [2,8]. But in all these cases time complexity per iteration are high and the accuracy of the relevant images are low.…”