Search

The objective of this paper was the development of a content-based image retrieval system, using siamese and triplet convolutional neural networks. These networks were used to generate visual descriptors, extracting semantic information from two images (siamese) or three images (triplet) at the same time. Then, a similarity learning was done, encoding these two or three visual descriptors. In the proposed scheme the storage of descriptors is not required. The experimental results show that the schemes based on convolutional neural networks extract more semantic information. The siamese and triplet architectures, apart from extracting semantic information, improved the image retrieval rate. It is concluded that the proposed scheme solved three of the main challenges in these systems, such as, semantic gap, similarity learning and storage space, which have not been solved in the previous works. [ABSTRACT FROM AUTHOR]

This paper addresses web analytics from the point of view of images, using big data technologies. Images are increasingly present on the web, and therefore any web analysis must consider them. The huge volume of available images requires the use of large computation infrastructures in order to generate this kind of analysis, as well as specific computer vision algorithms. Big data technologies are discussed that can be used for large-scale image processing. In addition, a novel distributed algorithm is proposed that can efficiently retrieve images from an online collection at low computational cost. This algorithm can be used in web analysis or future research requiring detailed image study, without requiring any special hardware. [ABSTRACT FROM AUTHOR]

The advent of the internet, the impact of digital photography and the ability to scan large volumes of images has revolutionized the stock photography agencies. The transformation from analogue to digital photography has involved a substantial change in daily work processes, providing a large number of images online. The present work will reveal the profile and the basic features that any stock photography agency must offer on issues such as image retrieval, on-screen visualization, and storage for many diverse editorial activities. [ABSTRACT FROM AUTHOR]

In Information Retrieval Systems, Image Retrieval Systems are located; they allow generating searching and storing processes of resources through coincidences by key words or other real-time methods. This type of systems can use visual content of images (color, texture, and brightness) as indices. In addition, these systems combine different attributes to improve the processes of classification and relevance of the searching process results known as "Information Quality." This paper presents the modeling of a Image Retrieval System which combines attributes of the contentbased retrieval such as color, texture, and shape, with the Information Quality-based retrieval as updating portability and relevance frequency in the context of Digital Image Collection of Aquatic Amazon Ecosystems of Universidad del Amazonia CAPREA Research Group. [ABSTRACT FROM AUTHOR]

The rise of internet has prompted image distribution agencies to market their stocks of images over the network. The present article evaluates existing systems for managing both image retrieval and copyright control that have been developed or applied by the major digital image banks. An assessment is also offered of the situation in Spain. In the discussion of image management systems, criteria are applied that relate to the conceptual analysis as well as to the inherent properties of the images. [ABSTRACT FROM AUTHOR]