In this paper, a novel low-level Image feature extraction and indexing scheme based on structure-texture Image decomposition is presented. The main idea of this work is to decompose database Images to structure and texture sub-Images to decrease the destructive effects of simultaneous existence of structure and texture information in the Image in indexing phase. It is also shown that precision in a typical content-based Image retrieval system can considerably increase by combining the feature vectors extracted from structure and texture sub-Images. An Image database containing 10000 Images of 82 different semantic groups is used to evaluate the proposed method. The results confirm the effectiveness of this method.

Several semantic Image search schemes have been recently proposed to retrieve Images from the web. However, the query context is regularly ignored in these techniques and hence, many of the returned Images are not adequately relevant. In this paper, we make use of context to further confine the outcome of the semantic search engines. For this purpose, we propose a hybrid search engine which utilizes concept and context for retrieving precise results. In the proposed model, an ontology is exploited for annotating Images and accomplishing search process in the semantic level. Furthermore, the query of the user is modified with the concepts available in the ontology. Next, we make use of search context of the user and augment the query with the information extracted from the user’s context to additionally eliminate irrelevant results. Experimental results show that the combination of concept and context is effective in retrieving and presenting the most relevant results to the user.

In this study, an Image retrieval system is proposed based on complex network model. Assuming a prior Image categorization, firstly, a multilayered complex network is constructed between the Images of each category according to the color, texture, and shape features. Secondly, by defining a meta-path as the way of connecting two Images in the network, a set of informative meta-paths are composed to find the similar Images by exploring the network. The established complex network provides an efficient way to benefit from the Image correlations to enhance the similarity search of the Images. On the other hand, employing diverse meta-paths with different semantics leads to measuring the Image similarities based on effective Image features for each category. The primary results indicate the efficiency and validity of the proposed approach.

E-commerce plays an important role in the world economy. A wide variety of websites have been designed to provide the ability of searching different types of products. Carpet is such a product which cannot be addressed easily with a special code in markets due to the huge variety in its specifications such as layout, color, and texture. This paper introduces a content-based Image retrieval system for carpet e-commerce application. This system helps development of the carpet e-commerce where an Image can be used instead of any tags including codes or models. An Image database containing various Persian carpet Images is also made for this application. Furthermore, several content-based Image retrieval methods are studied and applied on the carpet database and inspiring by the evaluation results, two methods, QCLD and DCDIP are proposed for carpet e-commerce application. Simulation results show 3. 1% and 2. 3% decrease on the ANMRR value for the proposed QCLD and DCDIP methods respectively. retrieval running times also are reported 2. 84 and 8. 15 seconds for the QCLD and DCDIP methods. In overall, these results reflect higher retrieval performance for the proposed methods.

Image retrieval is a basic task in many content-based Image systems. Achieving a high precision, while maintaining the computation time, is very important in relevance feedback-based Image retrieval systems. This paper establishes an analogy between this and the task of Image classification. Therefore, in the Image retrieval problem, we will obtain an optimized decision surface that separates the dataset Images into two categories of relevant/irrelevant Images corresponding to the query Image. This problem is viewed and solved as an optimization problem using the particle optimization algorithm. Although the particle swarm optimization (PSO) algorithm is widely used in the field of Image retrieval, no one uses it for a direct feature weighting. The information extracted from the user feedbacks will guide particles in order to find the optimal weights of various features of Images (color-, shape-or texture-based features). Fusion of these very non-homogenous features require a feature weighting algorithm that will take place by the help of the PSO algorithm. Accordingly, an innovative fitness function is proposed to evaluate each particle’, s position. The experimental results on the Wang dataset and Corel-10k indicate that the average precision of the proposed method is higher than the other semi-automatic and automatic approaches. Moreover, the proposed method suggests a reduction in the computational complexity in comparison with the other PSO-based Image retrieval methods.