A Novel Scheme for Content-Based Image Retrieval

李学龙,刘政凯,俞能海

DOI: https://doi.org/10.3321/j.issn:1000-436x.2002.12.012

2002-01-01

Abstract:This paper presents a new scheme for content-based image retrieval. A new similarity measurement method is proposed and a novel texture feature is applied on the test platform of iPhoto. The groundtruth is made of 56,600 images, and the result outperforms the traditional schemes.

What problem does this paper attempt to address?

SIFT-Based Image Retrieval Combining the Distance Measure of Global Image and Sub-Image

Bin Li,Xiangwei Kong,Zhe Wang,Haiyan Fu

DOI: https://doi.org/10.1109/IIH-MSP.2009.180

2009-01-01

Abstract:This paper presents a similarity match method based on global image and local sub-image using the SIFT features of digital images, and applies our algorithm to Content-Based Image Retrieval. In order to make the SIFT-based image retrieval results better, the most fundamental improvement comes in two areas. One is the introduction of the distance between the matched keypoints, and the shorter the distance between the matched keypoints, the lower the similarity measure. The other is that the image is partitioned off into sub-images, which reduces the mismatched keypoints. Experiments demonstrate effectiveness of the proposed approach compared with the traditional SIFT-based image retrieval and reveal it as a good option to image retrieval.
Content-Based Image Retrieval By Dominant Color Feature

袁昕,朱淼良,王东辉

DOI: https://doi.org/10.3321/j.issn:1003-9775.2002.11.013

2002-01-01

Abstract:Color is a kind of significant features in image retrieval. A new kind of image feature, namely Dominant Color Moment (DCM), which reflects spatial information of Dominant Color (DC), is taken into account. A feature model of DC features is also established. Several methods of similarity measurement based on the improvement of quadric form distance function are presented and compared. One of these methods, DCME, shows good effect in the CBIR experimental system implemented by means of WWW distribution.
A CONTENT-BASED IMAGE RETRIEVAL SCHEME IN JPEG COMPRESSED DOMAIN

Zhe-Ming Lu,Hans Burkhardt

2006-01-01

Abstract:Nowadays, a large number of images are compressed in JPEG (Joint Photo- graphic Experts Group) format. Therefore, content-based image retrieval (CBIR) for the JPEG images has attracted many people's attention and a series of algorithms directly based on the discrete cosine transform (DCT) domain have been proposed. However, the existing methods are far from the practical application. Thus, in this paper, a new image retrieval scheme for JPEG formatted images is presented. The color, spatial and frequency (texture) features based on the DCT domain are extracted for the later image retrieval. It doesn't require decompressing the images but directly retrieving in the DCT domain. Thus, compared with the spatial domain based retrieval methods for JPEG im- ages, the computation complexity can be greatly reduced. In addition, this retrieval system is suitable for all color images with different sizes. Experimental results demonstrate the advantages of the proposed retrieval scheme.
A Set of Novel Textural Features Based on 3D Co-Occurrence Matrix for Content-Based Image Retrieval

DC Tao,XL Li,YA Yuan,NH Yu,ZK Liu,XO Tang

DOI: https://doi.org/10.1109/icif.2002.1020979

2002-01-01

Abstract:This paper presents a set of novel texture features for content-based image retrieval (CBIR). CBIR require new algorithms for the automated extraction and indexing of salient image features, while texture features provide one important cue for the visual perception and discrimination of image content, so they are always one of the most popular representations in CBIR. The proposed features are based on the 3D co-occurrence matrix, and preliminary results show they are much better than traditional texture features. In this paper, experiments are made on an image database with 25000 images.
Design and Implementation on Content-based Web Image Retrieval System

LIAN Liping,FU Hongliang,CHEN Zhiqiang

2010-01-01

Abstract:Content-based image retrieval are an image information query retrieval which mainly based on color feature,texture feature,shape feature and so on.This paper firstly describes the image retrieval system architecture,focused on feature extraction technology and spider technology,a Web image retrieval system using combined color and texture features is also given and compared with which using single color feature aiming at concrete database,and discussions are done about the experiment results.The results show that the image retrieval system is effective and the sort is more reasonable.
Fast Content-Based Image Retrieval Based on Equal-Average K-Nearest-Neighbor Search Schemes

Zhe-Ming Lu,Hans Burkhardt,Sebastian Boehmer

DOI: https://doi.org/10.1007/11922162_20

2006-01-01

Abstract:The four most important issues in content-based image retrieval (CBIR) are how to extract features from an image, how to represent these features, how to search the images similar to the query image based on these features as fast as we can and how to perform relevance feedback. This paper mainly concerns the third problem. The traditional features such as color, shape and texture are extracted offline from all images in the database to compose a feature database, each element being a feature vector. The "linear scaling to unit variance" normalization method is used to equalize each dimension of the feature vector. A fast search method named equal-average K nearest neighbor search (EKNNS) is then used to find the first K nearest neighbors of the query feature vector as soon as possible based on the squared Euclidean distortion measure. Experimental results show that the proposed retrieval method can largely speed up the retrieval process, especially for large database and high feature vector dimension.
A novel learning-based method for image retrieval

Ce Gao,Yixu Song,Peifa Jia

2011-01-01

Journal of Computational Information Systems

Abstract:This paper provides a novel learning-based image retrieval algorithm by using affine invariant features. The traditional methods are based on color, texture, shape analysis and complicated descriptors. In this paper, we try to solve the problem in a probabilistic model. Based on machine learning, our method can recognize the query image in the database accurately and efficiently. We also try to improve the model in three levels. In the training level, we use Adaboost strategy to improve the classification ability of model. In the feature level, we reform the features to improve their distinguishing ability. In the Matching level, the keypoints are not independent any more. We use the structure information to constrain the matching, thus it can yield a better result. The experiment result demonstrates that our method can provide robust and powerful searching ability. At the same time it is very fast at run-time. © 2011 by Binary Information Press.
Deep Learning for Content-Based Image Retrieval: A Comprehensive Study

Ji Wan,Dayong Wang,Steven C. H. Hoi,Pengcheng Wu,Jianke Zhu,Yongdong Zhang,Jintao Li

DOI: https://doi.org/10.1145/2647868.2654948

2014-01-01

Abstract:Learning effective feature representations and similarity measures are crucial to the retrieval performance of a content-based image retrieval (CBIR) system. Despite extensive research efforts for decades, it remains one of the most challenging open problems that considerably hinders the successes of real-world CBIR systems. The key challenge has been attributed to the well-known ``semantic gap'' issue that exists between low-level image pixels captured by machines and high-level semantic concepts perceived by human. Among various techniques, machine learning has been actively investigated as a possible direction to bridge the semantic gap in the long term. Inspired by recent successes of deep learning techniques for computer vision and other applications, in this paper, we attempt to address an open problem: if deep learning is a hope for bridging the semantic gap in CBIR and how much improvements in CBIR tasks can be achieved by exploring the state-of-the-art deep learning techniques for learning feature representations and similarity measures. Specifically, we investigate a framework of deep learning with application to CBIR tasks with an extensive set of empirical studies by examining a state-of-the-art deep learning method (Convolutional Neural Networks) for CBIR tasks under varied settings. From our empirical studies, we find some encouraging results and summarize some important insights for future research.
Texture Content-Based Image Retrieval for Necktie Pattern

Chun-ying Liu,Zhi-geng Pan,Jin-xiang Dong

DOI: https://doi.org/10.1109/NCM.2009.160

2009-01-01

Abstract:By analyzing the character of necktie pattern and integrating image retrieval technology, a new approach to texture content-based image retrieval for necktie pattern is proposed in this paper. On the base of pattern extraction by color separation, Gabor wavelet transformation is used to extract pattern's eigenvector, an improved multiple-angle rotate-invariant similarity definition is presented. The experiment results are satisfying.
An Image Retrieval Method Based on Color and Texture Features

Leibo Liu,Weijie,Hu,Jian-ling,Xu,Chengliang

2006-01-01

Abstract:The technique of image retrieval is widely used in science experiment, military affairs, public security,advertisement, family entertainment, library and so on. The existing algorithms are mostly based on the characteristics of color, texture, shape and space relationship. This paper introduced an image retrieval algorithm, which is based on the matching of weighted EMD(Earth Mover's Distance) distance and texture distance. EMD distance is the distance between the histograms of two images in HSV(Hue, Saturation, Value) color space, and texture distance is the L1 distance between the texture spectra of two images. The experimental results show that the retrieval rate can be increased obviously by using the proposed algorithm.
Image Retrieval Method Based Texture Watermarks

M Li,Zm Lu,Yl Qiao

2005-01-01

Abstract:An image retrieval method using watermarking is proposed in this paper This method consists of three parts: computing texture features (CTF), embedding these texture features in the images (ETI) and improved image retrieval method (IIR). We choose three texture features in CTF They are texture browsing descriptor, HTD (homogeneous texture descriptor) and Edge histogram descriptor. Then these features are invisibly embedded into images. We embed watermark in DCT domain using dither modulation algorithm in ETI. Then we use IIR to retrieval images. Experimental results show the effectiveness of the proposed scheme.
Indexing of Content-Based Image Retrieval System with Image Understanding Approach

李学龙,刘政凯,俞能海

2003-01-01

Abstract:This paper presents a novel efficient semantic image classification algorithm for high-level feature indexing of high-dimension image database. Experiments show that the algorithm performs well. The size of the train set and the test set is 7 537 and 5 000 respectively. Based on this theory, another ground is built with 12,000 images, which are divided into three classes: city, landscape and person, the total result of the classifications is 88.92%, meanwhile, some preliminary results are presented for image understanding based on semantic image classification and low level features. The groundtruth for the experiments is built with the images from Corel database, photos and some famous face databases.
Image Interactive Retrieval Model Based on Integrated Region Similarity

Chen Tian,Ren Zhi

2003-01-01

Abstract:Efficient content based image retrieval has developed tremendously in many application areas. Content based image retrieval research areas had established many systems, but these systems have deficiency for actually usage:(1) these systems are expected to process different kinds of image retrieval by same method and (2) these systems are designed without considering user requirement. In fact, different kinds of image retrieval need different retrieval mode. So it presents image interactive retrieval based on integrated region similarity match. This system is an image retrieval system with interaction and semantic classification of based on wavelet feature extraction and image integrated region similarity method. Compared with other retrieval methods of based on image content, the method enhances retrieval efficiency by permitting adaptive search and interaction and narrowing search area. Experiment result shows that the system is more precise and efficient than other retrieval systems.
New Image Retrieval Approach Based On Interest Points

Jw Han,G Lei

DOI: https://doi.org/10.1117/12.473035

2002-01-01

Abstract:Content-based image retrieval (CBIR) is becoming a hot research point in the field of multimedia information retrieval. Interest points are local features with high informational content. So, this paper proposes a novel method for image retrieval using interest points, which contains three key stages: interest points detection, image features description based on interest points and similarity measure between two images. In the process of detecting interest points, firstly, we use a self-adaptive filter to smooth image, and then use detector to find interest points. In the stage of image features description, we design a 3D histogram to represent image, which contains local gray changes of interest points, mutual position relations among interest points and interest points distribution of the whole image. In the stage of similarity measure, we use the distance between two 3D histograms to calculate similarity between two images. Lots of experimental results based on a database containing 1500 images demonstrate our proposed approach is efficient.
Combining similarity measures in content-based image retrieval guided by mutual information

Zhi Lijia,Zhang Shaomin,Zhao Dazhe,Zhao Hong,Lin Shukuan

2011-01-01

Journal of Image and Graphics

Abstract:The lack of accordance between the information that one can extract from an image and the interpretability of the same image in a given situation is the most important factor that hampers the accuracy of content-based image retrieval(CBIR).Recently,the combination of several similarity measures draws much interest in the CBIR area,It can be shown that is effective in reducing this discordance.The core problem is:how to choose a better way to combine these similarities? In this paper,we propose a new combination algorithm.It combines similarity measures under the sum rule based on mutual information which estimates the correlation between the continuous random variable similarity measures and the discrete random variable similarity.The experimental results show that this algorithm achieves a high accuracy and efficiency in real-world image collections.
A parameterized evaluation model of similarity on multi-target retrieval for remote sensing image

杜震洪,刘仁义,周永福,曾志

DOI: https://doi.org/10.6046/gtzyyg.2015.04.02

2015-01-01

Abstract:In content - based image retrieval ( CBIR ) , the feature selection of target retrieval and similarity computing is a hot issue of current research. As the object-oriented image processing methods are conducive to describing the spatial relationships between objects, the authors proposed an parameterized evaluation model of similarity for multi-target retrieval on the basis of an analysis of multi-feature descriptor. This model not only takes into account the global spatial relationship between image objects but also considers the local features of image targets. The model organizes the multi -dimensional features for an image using the analytic hierarchy process ( AHP) method under moving picture expert group ( MPEG7 ) standard, then the eigenvalues are calculated using existing methods, and finally the weighting coefficients are set for calculating the overall similarity on the basis of user preferences so as to evaluate the effect of multi-target retrieval. Thus it can provide a powerful reference and experience for multi-target retrieval in an image database. The experimental results show that target retrieval is not only related to the method of similarity calculation but also related to the complexity of the image itself.
A Rich Get Richer Strategy For Content-Based Image Retrieval

Lj Duan,W Gao,Jy Ma

DOI: https://doi.org/10.1007/3-540-40053-2_26

2000-01-01

Abstract:A novel relevance feedback approach to image retrieval, Rich Get Richer (RGR) Strategy, is proposed in this paper. It is based on the general framework of Bayesian inference in statistics. The user's feedback information is propagated into the retrieval process step by step. The more promising images are emphasized by the Rich Get Richer (RGR) Strategy. On the contrary, the less promising ones are de-emphasized. The experimental results show that the proposed approach can capture the user's information need more precisely. By using RGR, the average precision improves from 5 to 20% for each interaction.
A Novel Region-Based Image Retrieval Method Using Relevance Feedback

Feng Jing,Bo Zhang,Fuzong Lin,Wei-Ying Ma,HongJiang Zhang

DOI: https://doi.org/10.1145/500933.500943

2001-01-01

Abstract:Content-based image retrieval using region segmentation has been an active research area in the past few years. Constrasting to traditional approaches, which compute only global features of images, the region-based methods extract features of the segmented regions and perform similarity comparisons at the granularity of region. In this paper, we propose a novel region-based retrieval method, Self-Learned Region Importance (SLRI). In this method, image similarity measure is based on the region importance learned from users' feedback. The region importance that coincides that human perception con not only be used in a query session, but also be memorized and cumulated for future queries. Experimental results on a database of about 8,600 general-purposed images show the effectiveness of our method using relevance feedback.
Ranking consistency for image matching and object retrieval

Yanzhi Chen,Xi Li,Anthony Dick,Rhys Hill

DOI: https://doi.org/10.1016/j.patcog.2013.09.011

IF: 8

2014-01-01

Pattern Recognition

Abstract:The goal of object retrieval is to rank a set of images by the similarity of their contents to those of a query image. However, it is difficult to measure image content similarity due to visual changes caused by varying viewpoint and environment. In this paper, we propose a simple, efficient method to more effectively measure content similarity from image measurements. Our method is based on the ranking information available from existing retrieval systems. We observe that images within the set which, when used as queries, yield similar ranking lists are likely to be relevant to each other and vice versa. In our method, ranking consistency is used as a verification method to efficiently refine an existing ranking list, in much the same fashion that spatial verification is employed. The efficiency of our method is achieved by a list-wise min-Hash scheme, which allows rapid calculation of an approximate similarity ranking. Experimental results demonstrate the effectiveness of the proposed framework and its applications.
A Hierarchical Characterization Scheme for Image Retrieval

WY Liu,T Wang,HJ Zhang

DOI: https://doi.org/10.1109/icip.2000.899281

2000-01-01

Abstract:In this paper, we present a study of hierarchically characterizing image content from coarse level to fine level conducted using a series of shape features as a case in point. In this work, we have empirically proved that this hierarchical characterization is effective and necessary. We have used this hierarchical multimedia content characterization scheme in our shape retrieval experiments and see that the hierarchical method obtains similar performance in the retrieval of similar shapes to that of the simultaneous method (calculating similarity based on all features at a time), while the efficiency has been increased significantly

A Novel Scheme for Content-Based Image Retrieval

SIFT-Based Image Retrieval Combining the Distance Measure of Global Image and Sub-Image

Content-Based Image Retrieval By Dominant Color Feature

A CONTENT-BASED IMAGE RETRIEVAL SCHEME IN JPEG COMPRESSED DOMAIN

A Set of Novel Textural Features Based on 3D Co-Occurrence Matrix for Content-Based Image Retrieval

Design and Implementation on Content-based Web Image Retrieval System

Fast Content-Based Image Retrieval Based on Equal-Average K-Nearest-Neighbor Search Schemes

A novel learning-based method for image retrieval

Deep Learning for Content-Based Image Retrieval: A Comprehensive Study

Texture Content-Based Image Retrieval for Necktie Pattern

An Image Retrieval Method Based on Color and Texture Features

Image Retrieval Method Based Texture Watermarks

Indexing of Content-Based Image Retrieval System with Image Understanding Approach

Image Interactive Retrieval Model Based on Integrated Region Similarity

New Image Retrieval Approach Based On Interest Points

Combining similarity measures in content-based image retrieval guided by mutual information

A parameterized evaluation model of similarity on multi-target retrieval for remote sensing image

A Rich Get Richer Strategy For Content-Based Image Retrieval

A Novel Region-Based Image Retrieval Method Using Relevance Feedback

Ranking consistency for image matching and object retrieval

A Hierarchical Characterization Scheme for Image Retrieval