적합성 피드백을 통해 결정된 가중치를 갖는 시각적 특성에 기반을 둔 이미지 검색 모델

송지영(Ji-Young Song),김우철(Woo-Cheol Kim),김승우(Seung-Woo Kim),박상현(Sanghyun Park) 한국정보과학회 2007 정보과학회논문지 : 데이타베이스 Vol.34 No.3

디지털 이미지의 양이 증가함에 따라 원하는 이미지를 정확하고 빠르게 찾을 수 있는 방법의 필요성이 증가하고 있다. 이미지 검색 방법으로는 이미지의 색상이나 명암과 같은 시각적 특성을 검색 조건으로 이용하는 내용 기반 검색과 이미지를 설명하는 키워드를 검색 조건으로 이용하는 키워드 기반 검색이 있다. 하지만 이러한 방법만으로는 사용자가 원하는 이미지를 정확하게 찾기 힘들다는 문제점이 제기되어 왔다. 따라서 최근에는 검색 도중 사용자의 응답을 받아 사용자의 요구를 파악함으로써 향상된 검색결과를 제공하는 적합성 피드백에 대한 연구가 많이 진행되고 있다. 하지만 적합성 피드백을 이용하는 방법들도 원하는 결과를 얻기 위해서는 여러 번의 피드백을 필요로 하고 질의 수행이 완료된 후에는 얻어진 피드백 정보를 재사용하지 못한다는 단점이 있다. 따라서 본 논문에서는 이미지에 키워드를 연결한 후 사용자의 피드백 정보를 반영하여 키워드의 신뢰도를 조절함으로써 키워드 기반 이미지 검색의 정확도를 높일 수 있는 모델을 제안한다. 제안된 모델에서는 사용자로부터 피드백을 받은 이미지뿐만 아니라 긍정적 피드백을 받은 이미지들이 공통적으로 가지는 시각적 특성과 유사한 시각적 특성을 가지는 다른 이미지들까지도 키워드의 신뢰도를 조정함으로써 좀 더 빠른 시간 내에 검색 결과의 정확도를 높이도록 한다. 제안한 방법의 정확성을 검증하기 위한 실험 결과에 따르면, 같은 횟수의 피드백을 받으면서도 재현율과 정확률은 빠른 증가를 보이는 것으로 나타났다. Increasing amount of digital images requires more accurate and faster way of image retrieval. So far, image retrieval method includes content-based retrieval and keyword based retrieval, the former utilizing visual features such as color and brightness and the latter utilizing keywords which describe the image. However, the effectiveness of these methods as to providing the exact images the user wanted has been under question. Hence, many researchers have been working on relevance feedback, a process in which responses from the user are given as a feedback during the retrieval session in order to define user’s need and provide improved result. Yet, the methods which have employed relevance feedback also have drawbacks since several feedbacks are necessary to have appropriate result and the feedback information can not be reused. In this paper, a novel retrieval model has been proposed which annotates an image with a keyword and modifies the confidence level of the keyword in response to the user’s feedback. In the proposed model, not only the images which have received positive feedback but also the other images with the visual features similar to the features used to distinguish the positive image are subjected to confidence modification. This enables modifying large amount of images with only a few feedbacks ultimately leading to faster and more accurate retrieval result. An experiment has been performed to verify the effectiveness of the proposed model and the result has demonstrated rapid increase in recall and precision while receiving the same number of feedbacks.

Image Retrieval Process Based on Relevance Feedback and Ontology Using Decision Tree

Debnath Bhattacharyya,Dipankar Hazra,Tai-hoon Kim 보안공학연구지원센터 2015 International Journal of Multimedia and Ubiquitous Vol.10 No.10

In this paper, another strategy for immediate features based image recovery is proposed. Image database is developed with low level texture features got from Gray Level Co- Occurrence Matrix (GLCM) and measurable techniques for Tamura. Semantic level inquiries from the user mapped to the low level peculiarities at recovery time to recover the required images. Images with more than one moderate features can be recovered by utilizing intersection of images recovered by each of the queried feature. Artificial Neural Network (ANN) is utilized as a part of the following steps in the wake of accepting user inputs. In spite of the fact that semantics are utilized as search key as a part of the beginning steps, low level features are utilized as a part of the ANN based searching in later steps. Back propagation Algorithm is utilized as a part of learning step. This ANN based relevance feedback technique enhances accuracy of immediate feature based image retrieval method. Decision tree (DT) can likewise be connected in relevance feedback stage. Decision tree is framed in training stage and images will be tested by of the decision tree. Relation storing ontology related information is utilized as a part of every phase of retrieval procedure to evacuate ambiguities identified with synonyms and hypernym-homonym sets.

Medical Image Retrieval with Relevance Feedback via Pairwise Constraint Propagation

( Menglin Wu ),( Qiang Chen ),( Quansen Sun ) 한국인터넷정보학회 2014 KSII Transactions on Internet and Information Syst Vol.8 No.1

Relevance feedback is an effective tool to bridge the gap between superficial image contents and medically-relevant sense in content-based medical image retrieval. In this paper, we propose an interactive medical image search framework based on pairwise constraint propagation. The basic idea is to obtain pairwise constraints from user feedback and propagate them to the entire image set to reconstruct the similarity matrix, and then rank medical images on this new manifold. In contrast to most of the algorithms that only concern manifold structure, the proposed method integrates pairwise constraint information in a feedback procedure and resolves the small sample size and the asymmetrical training typically in relevance feedback. We also introduce a long-term feedback strategy for our retrieval tasks. Experiments on two medical image datasets indicate the proposed approach can significantly improve the performance of medical image retrieval. The experiments also indicate that the proposed approach outperforms previous relevance feedback models.

기억력을 가진 적합성 피드백

오상욱,정민교 서울여자대학교 컴퓨터과학연구소 2004 정보기술논문지 Vol.3 No.-

내용기반 영상검색(content-based image retrieval) 시스템은 영상의 시각적인 특징정보(색깔, 텍스처, 모양 등)를 사용하여 질의 영상(query image)과 유사한 영상을 검색하게 된다. 그러나 이런 하위 레벨의 시각적인 특징정보만으로는, 영상에 내재된 상위 레벨의 의미 있는 정보(예를 들면, 객체, 이벤트, 상호관계 정보 등)를 제대로 표현하지 못하게 된다. 그래서 내용기반 영상검색 시스템의 이러한 단점을 보완하는 한 방법으로, 적합성 피드백(RF: Relevance Feedback) 개념을 이용한 영상검색 기법이 일찍부터 사용되었다. 그러나 기존 RF 기법에서는 사용자로부터 어렵게 획득한 적합성 정보를 검색 종료 후 막연히 폐기하는 단점을 가지고 있다. 본 논문에서는 이런 사용자의 적합성 정보를 버리지 않고, 체계적으로 저장하여 영상의 검색 효율을 높이는 새로운 개념의 RF 기법을 제안한다. 새로 제안된 RF는 시간의 흐름에 따라 축적되어 저장된 상위 레벨의 적합성 정보를, 하위 레벨의 시각정보와 동적으로 결합하여 사용함으로써 검색의 효율성을 크게 향상시킨다. Content-based image retrieval systems acquire the visual content of an image such as color, texture, shape, etc. and use these low-level visual features to find all the images similar to a query image. Unfortunately, the low-level visual content alone cannot capture high-level image semantics(for example, objects, events, and relationships) in a meaningful way. To get around those difficulties in the content-based retrieval systems, an image retrieval method based on the concept of relevance feedback(RF) has been used earlier, which gradually refines a query by the relevance information from a user. However, the traditional RF method has a drawback to simply throw away the user’s feedback information as soon as a search session ends. In this paper, a new version of RF is proposed, which systematically accumulates human perceptual responses over time through relevance feedback mechanism. Specifically, the newly proposed RF combines the accumulated high-level relevance information with low-level visual features to further improve the retrieval effectiveness.

An Effective Relevance Feedbackbased Image Retrieval using Color and Texture

Jung, Sung-Hwan Korea Multimedia Society 2003 멀티미디어학회논문지 Vol.6 No.4

In this paper, we proposed an image retrieval system with a simple and effective relevance feedback, called RAP(Reward and Punishment) algorithm. First, color and texture features were extracted from the images. Next, the extracted feature values were used for image retrieval in various forms. We applied the relevance feedback to the initial retrieved images from the image retrieval system, and compared its result with that of the conventional system. In the experiment using the test image database of 16 class 512 images, the proposed system showed the better retrieval performance of about 10∼l7 % than that of the conventional INRIA system in each relevance feedback step.

Genetic Algorithm based Relevance Feedback for Content-based Image Retrieval

서광규 한국반도체디스플레이기술학회 2008 반도체디스플레이기술학회지 Vol.7 No.4

This paper explores a content-based image retrieval framework with relevance feedback based on genetic algorithm (GA). This framework adopts GA to learn the user preferences using the similarity functions defined for all available descriptors. The objective of the GA-based learning methods is to learn the user preferences using the similarity functions and to find a descriptor combination function that best represents the user perception. Experiments were performed to validate the proposed frameworks. The experiments employed the natural image databases and color and texture descriptors to represent the content of database images. The proposed frameworks were compared with the other two relevance feedback methods regarding effectiveness in image retrieval tasks. Experiment results demonstrate the superiority of the proposed method.

Genetic Algorithm based Relevance Feedback for Content-based Image Retrieval

Seo, Kwang-Kyu The Korean Society Of Semiconductor Display Techno 2008 반도체디스플레이기술학회지 Vol.7 No.4

This paper explores a content-based image retrieval framework with relevance feedback based on genetic algorithm (GA). This framework adopts GA to learn the user preferences using the similarity functions defined for all available descriptors. The objective of the GA-based learning methods is to learn the user preferences using the similarity functions and to find a descriptor combination function that best represents the user perception. Experiments were performed to validate the proposed frameworks. The experiments employed the natural image databases and color and texture descriptors to represent the content of database images. The proposed frameworks were compared with the other two relevance feedback methods regarding effectiveness in image retrieval tasks. Experiment results demonstrate the superiority of the proposed method.

Support Vector Machine Learning for Region-Based Image Retrieval with Relevance Feedback

김덕환,Jae-Won Song,Ju-Hong Lee,Bum-Ghi Choi 한국전자통신연구원 2007 ETRI Journal Vol.29 No.5

We present a relevance feedback approach based on multi-class support vector machine (SVM) learning and cluster-merging which can significantly improve the retrieval performance in region-based image retrieval. Semantically relevant images may exhibit various visual characteristics and may be scattered in several classes in the feature space due to the semantic gap between low-level features and high-level semantics in the user's mind. To find the semantic classes through relevance feedback, the proposed method reduces the burden of completely re-clustering the classes at iterations and classifies multiple classes. Experimental results show that the proposed method is more effective and efficient than the two-class SVM and multi-class relevance feedback methods.

An Approach for the Cross Modality Content-Based Image Retrieval between Different Image Modalities

Jeong, Inseong,Kim, Gihong Korean Society of Surveying 2013 한국측량학회지 Vol.31 No.6

CBIR is an effective tool to search and extract image contents in a large remote sensing image database queried by an operator or end user. However, as imaging principles are different by sensors, their visual representation thus varies among image modality type. Considering images of various modalities archived in the database, image modality difference has to be tackled for the successful CBIR implementation. However, this topic has been seldom dealt with and thus still poses a practical challenge. This study suggests a cross modality CBIR (termed as the CM-CBIR) method that transforms given query feature vector by a supervised procedure in order to link between modalities. This procedure leverages the skill of analyst in training steps after which the transformed query vector is created for the use of searching in target images with different modalities. Current initial results show the potential of the proposed CM-CBIR method by delivering the image content of interest from different modality images. Despite its retrieval capability is outperformed by that of same modality CBIR (abbreviated as the SM-CBIR), the lack of retrieval performance can be compensated by employing the user's relevancy feedback, a conventional technique for retrieval enhancement.

An Approach for the Cross Modality Content-Based Image Retrieval between Different Image Modalities

정인성,김기홍 한국측량학회 2013 한국측량학회지 Vol.31 No.6

CBIR is an effective tool to search and extract image contents in a large remote sensing image databasequeried by an operator or end user. However, as imaging principles are different by sensors, their visualrepresentation thus varies among image modality type. Considering images of various modalities archived inthe database, image modality difference has to be tackled for the successful CBIR implementation. However,this topic has been seldom dealt with and thus still poses a practical challenge. This study suggests a crossmodality CBIR (termed as the CM-CBIR) method that transforms given query feature vector by a supervisedprocedure in order to link between modalities. This procedure leverages the skill of analyst in training stepsafter which the transformed query vector is created for the use of searching in target images with differentmodalities. Current initial results show the potential of the proposed CM-CBIR method by delivering the imagecontent of interest from different modality images. Despite its retrieval capability is outperformed by that ofsame modality CBIR (abbreviated as the SM-CBIR), the lack of retrieval performance can be compensated byemploying the user’s relevancy feedback, a conventional technique for retrieval enhancement.