Semi-Automatic Image Annotation

2010

Effective keyword search on image databases is a major open problem, due to the inherent imprecision of keywords (tags) used to describe images' content. In this paper we present a novel approach to deal with this problem, as implemented in the Scenique image retrieval and browsing system. Scenique is based on a multi-dimensional model, where each dimension is a tree-structured taxonomy of concepts, also called semantic tags, that are used to describe the content of images. We first describe an original algorithm, called Ostia (Optimal Semantic Tags for Image Annotation), that, by exploiting low-level visual features, tags, and metadata associated to an image, is able to predict a high-quality set of semantic tags for that image. Then, we describe how semantic tags can be effectively used for the purpose of improving the precision of keyword search.

2004

ABSTRACT Compared with content-based image retrieval, annotationbased image retrieval is more practical in some application domains. Users' information needs and the semantic contents of images can be represented by textual information more easily. We describe two problems which are unique to annotation-based image retrieval and would be worthy of further research. Contextual information embedded in data may be used to address these problems.

SPIE Proceedings, 1997

Content-Based Image Retrieval (CBIR) has become one of the most active research areas in the past few years. Many visual feature representations have been explored and many systems built. While these research e orts establish the basis of CBIR, the usefulness of the proposed approaches is limited. Speci cally, these e orts have relatively ignored two distinct characteristics of CBIR systems: (1) the gap between high level concepts and low l e v el features (2) subjectivity o f h uman perception of visual content. This paper proposes a relevance feedback based interactive retrieval approach, which e ectively takes into account the above t wo c haracteristics in CBIR. During the retrieval process, the user's high level query and perception subjectivity are captured by dynamically updated weights based on the user's relevance feedback. The experimental results show that the proposed approach greatly reduces the user's e ort of composing a query and captures the user's information need more precisely. 1 Introduction With the advances in the computer technologies and the advent o f the WorldWide Web, there has been an explosion in the amount and complexity of digital data being generated, stored, transmitted, analyzed, and accessed. Much of this information is multimedia in nature, including digital images, video, audio, graphics, and text data. In order to make use of this vast amount of data, e cient and e ective t e c hniques to retrieve m ultimedia information based on its content need to be developed. Among the various media types, images are of prime importance. Not only it is the most widely used media type besides text, but it is also the basis for representing and retrieving videos and other multimedia information. This paper deals with the retrieval of images based on their contents, even though the approach is readily generalizable to other media types. Keyword annotation is the traditional image retrieval paradigm. In this approach, the images are rst annotated manually by k eywords. They can then be retrieved by their corresponding annotations. However, there are three main di culties with this approach, i.e. the large amount of manual e ort required in developing the annotations, the di erences in interpretation of image contents, and inconsistency of the keyword assignments among di erent indexers 1, 2, 3]. As the size of image repositories increases, keyword annotation approach becomes infeasible.

The Latent Semantic Indexing (LSI) and Probabilistic Latent Semantic Indexing (pLSI) in the annotation based image retrieval (ABIR) task it gives better result. In the existing systems technical operations for online image retrieval are of i) automatic annotation ii) indexing iii) annotation based retrieval of images. The keyword space with the images and their annotation also the keyword relevance are captured in the AMC. The problems addressed with these are it brings large gap between the users query and the image description. It gives confused state and collaboration in selecting documents, also limitation in image description. Here to introduce the new method called Markovian chain with ranking method. A stochastic distance between images, based on their annotation and the keyword relevance captured in the AMC. By means of a new measure of Markovian state similarity, the mean first cross passage time (CPT), optimality properties of the proposed distance are proved. This method is shown to possess certain theoretical advantages. Optimal ranking is set through a counter for each customer response process.

In the recent years, multimedia content has grown increasingly over the Internet, especially in social networks, where users often post images using their mobile devices. In these networks such as Flickr, the content is later used in search operations when some users want to find something using a specific query. Nowadays, searching into these networks is primarily made using the title and the keywords associated to resources added by users that have posted the content. The problem we face comes from the fact that in many cases, the title or the related keywords are not relevant to the resource and only after we analyse the image, can we conclude what it contains in reality. The project that we want to present in this article proposes that each image is connected to relevant keywords according to its content. In order to do this, the first step was to create a collection of images that was annotated by human annotators, while the second step was to expand this collection of images performing search on the Internet using keywords associated to the initial collection of annotated images. Currently, for a new picture, we can identify similar images in our collection of images and based on the keywords associated with them, we can determine what keywords characterize this new image. The evaluation of this system has demonstrated that our approach works efficiently for images for which we can find similar images in our collection.