Content-based image retrieval by a semi-supervised Particle Swarm Optimization

Abstract

An innovative approach based on an evolutionary stochastic algorithm, namely the Particle Swarm Optimizer (PSO), is proposed in this paper as a solution to the problem of intelligent retrieval of images in large databases. The problem is recast to an optimization one, where a suitable cost function is minimized through a customized PSO. Accordingly, the relevance-feedback is used in order to exploit the information of the user with the aim of both guiding the particles inside the search space and dynamically assigning different weights to the features.

Visual search reranking via adaptive particle swarm optimization

“…Owing to its stochastic optimization in an exploration of the search space, PSO has been employed in many content-based image retrieval (CBIR) systems [20][21][22][23]. Okayama et al proposed to first collect user-preferred images in a preference feature space, and then optimize the parameters for feature weighting according to user's evaluation of the retrieved images by PSO [23].…”

“…Okayama et al proposed to first collect user-preferred images in a preference feature space, and then optimize the parameters for feature weighting according to user's evaluation of the retrieved images by PSO [23]. Broilo et al combined a relevance feedback (RF) approach with PSO for image retrieval [20][21][22]. PSO was used to dynamically update the feature space by optimally weighting the features based on user's feedback.…”

“…Starting from a diffuse population (or called a ''swarm''), the individuals (particles) move among the search space and ultimately cluster themselves in the region where the optima (food) locates. This mechanism has yet been applied to solve lots of optimization problems, such as system scheduling [13,14], automatic object tracking [15], signal processing [16], real time robot path planning and gesture recognition [17], image feature analysis [18], text processing [19], content-based image retrieval [20][21][22][23], and so on.…”

“…This is different from traditional content-based image retrieval through [20][21][22][23]. In visual search reranking, an initial ranked list of search results (usually from textbased search) are given, then the problem is to discover relevant information from the initial ranked list and rerank those documents for better search accuracy.…”

An Evolutionary Stochastic Approach for Efficient Image Retrieval using Modified Particle Swarm Optimization

“…A combined use of color and texture would provide better performance than that of color or texture alone [13] and the feature vector consists of the color and texture features [14]. Most of the image retrieval methods are not stochastic; consequently, searching in different solution space is not possible [15].…”