• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.711-716
• /
• 2006

Since the introduction of digital camera to the mass market, the number of digital photos owned by an individual is growing at an alarming rate. This phenomenon naturally leads to the issues of difficulties while searching and browsing in the personal digital photo archive. Traditional approach typically involves content-based image retrieval using computer vision algorithms. However, due to the performance limitations of these algorithms, at least on the casual digital photos taken by non-professional photographers, more recent approaches are centered on time-based clustering algorithms, analyzing the shot times of photos. These time-based clustering algorithms are based on the insight that when these photos are clustered according to the shot-time similarity, we have "event clusters" that will help the user browse through her photo archive. It is also reported that one of the remaining problems with the time-based approach is that people perceive events in different scales. In this paper, we present an adaptive time-based clustering algorithm that exploits the usage history of digital photos in order to infer the user's preference on the event granularity. Experiments show significant performance improvements in the clustering accuracy.

• Journal of KIISE:Software and Applications
• /
• v.31 no.9
• /
• pp.1161-1170
• /
• 2004

This paper proposes to segment skin color areas using a clustering algorithm. Most of previously proposed clustering algorithms have some difficulties, since they generally detect hyperspherical clusters, run in a batch mode, and predefine a number of clusters. In this paper, we use a well-known elliptical clustering algorithm, an EM algorithm, and modify it to learn on-line and find automatically the number of clusters, called to an EAM algorithm. The effectiveness of the EAM algorithm is demonstrated on a task of skin color region segmentation. Experimental results present the EAM algorithm automatically finds a right number of clusters in a given image without any information on the number. Comparing with the EM algorithm, we achieved better segmentation results with the EAM algorithm. Successful results were achieved to detect and segment skin color regions using a conditional probability on a region. Also, we applied to classify images with persons and got good classification results.

• Journal of Korea Multimedia Society
• /
• v.10 no.4
• /
• pp.441-449
• /
• 2007

In clustering for image segmentation, large amount of computation and typical segmentation errors have been important problems. In the paper, we suggest a new method for minimizing these problems. Markers in marker-controlled watershed transform represent segmented areas because they are starting-points of extending areas. Thus, clustering restricted by marker pixels can reduce computational complexity. In our proposed method, the markers are selected by Gabor texture energy, and cluster information of them are generated by FCM (fuzzy c-mean) clustering. Generated areas from watershed transform are merged by using cluster information of markers. In the test of Brodatz' texture images, we improved typical partition-errors obviously and obtained less computational complexity compared with previous FCM clustering algorithms. Overall, it also took regular computational time.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.40 no.1
• /
• pp.10-19
• /
• 2003

Caption information in news videos can be useful for video indexing and retrieval since it usually suggests or implies the contents of the video very well. In this paper, a new algorithm for extracting and recognizing characters from news video is proposed, without a priori knowledge such as font type, color, size of character. In the process of text region extraction, in order to improve the recognition rate for videos with complex background at low resolution, continuous frames with identical text regions are automatically detected to compose an average frame. The image of the averaged frame is projected to horizontal and vertical direction, and we apply region filling to remove backgrounds to produce the character. Then, K-means color clustering is applied to remove remaining backgrounds to produce the final text image. In the process of character recognition, simple features such as white run and zero-one transition from the center, are extracted from unknown characters. These feature are compared with the pre-composed character feature set to recognize the characters. Experimental results tested on various news videos show that the proposed method is superior in terms of caption extraction ability and character recognition rate.

• Journal of Korea Multimedia Society
• /
• v.15 no.9
• /
• pp.1149-1155
• /
• 2012

In this paper, we propose cause diagnosis method of semiconductor defects from semiconductor industrial images. Our method constructs feature database (DB) of defect images. Then, defect and input images are subdivided by uniform block. And the block similarity is measured using histogram kai-square distance after color histogram calculation. Then, searched blocks in each image are merged into connected objects using clustering. Finally, the most similar defect image from feature DB is searched with the defect cause by measuring cluster similarity based on features of each cluster. Our method was validated by calculating the search accuracy of n output images having high similarity. With n = 1, 2, 3, the search accuracy was measured to be 100% regardless of defect categories. Our method could be used for the industrial applications.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.43 no.4 s.310
• /
• pp.1-8
• /
• 2006

This paper proposes an improved FCM(Fuzzy C-means) algorithm using intercluster and entropy-based weight in gray image. The fuzzy clustering methods have been extensively used in the image segmentation since it extracts feature information of the region. Most of fuzzy clustering methods have used the FCM algorithm. But, FCM algorithm is still sensitive to noise, as it does not include spatial information. In addition, it can't correctly classify pixels according to the feature-based distributions of clusters. To solve these problems, we applied a weight and intercluster to the traditional FCM algorithm. A weight is obtained from the entropy information based on the cluster's number of neighboring pixels. And a membership for one pixel is given based on the information considering the feature-based intercluster. Experiments has confirmed that the proposed method was more tolerant to noise and superior to existing methods.

• The KIPS Transactions:PartB
• /
• v.12B no.7 s.103
• /
• pp.737-744
• /
• 2005

In this paper, we propose a three-level color clustering algerian for the binarization of text regions extracted from natural scene images. The proposed algorithm consists of three phases of color segmentation. First, the ordinary images in which the texts are well separated from the background, are binarized. Then, in the second phase, the input image is passed through a high pass filter to deal with those affected by natural or artificial light. Finally, the image Is passed through a low pass filter to deal with the texture in texts and/or background. We have shown that the proposed algorithm is more effective used gray-information binarization algorithm. To evaluate the effectiveness of the proposed algorithm we use a commercial OCR software ARMI 6.0 to observe the recognition accuracies on the binarized images. The experimental results on word and character recognition show that the proposed approach is more accurate than conventional methods by over $35\%$.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.12
• /
• pp.3315-3337
• /
• 2012

Crowded scenes analysis is a challenging topic in computer vision field. How to detect diverse motion patterns in crowded scenarios from videos is the critical yet hard part of this problem. In this paper, we propose a novel approach to mining motion patterns by utilizing motion information during both long-term period and short interval simultaneously. To capture long-term motions effectively, we introduce Motion History Image (MHI) representation to access to the global perspective about the crowd motion. The combination of MHI and optical flow, which is used to get instant motion information, gives rise to discriminative spatial-temporal motion features. Benefitting from the robustness and efficiency of the novel motion representation, the following motion pattern mining is implemented in a completely unsupervised way. The motion vectors are clustered hierarchically through automatic hierarchical clustering algorithm building on the basis of graphic model. This method overcomes the instability of optical flow in dealing with time continuity in crowded scenes. The results of clustering reveal the situations of motion pattern distribution in current crowded videos. To validate the performance of the proposed approach, we conduct experimental evaluations on some challenging videos including vehicles and pedestrians. The reliable detection results demonstrate the effectiveness of our approach.

• Journal of Internet Computing and Services
• /
• v.14 no.2
• /
• pp.1-13
• /
• 2013

Recent advance and popularization of smart devices and web application services based on cloud computing have made end-users to directly produce and, at the same time, consume the image contents. This leads to demands of unified contents management services. Thus, this paper proposestag clustering method based on semantic similarity for effective image categorization. We calculate the cost of semantic similarity between tags and cluster tags that are closely related. If tags are in a cluster, we suppose that images with them are also in a same cluster. Furthermore, we could recommend tags for new images on the basis of initial clusters.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.10
• /
• pp.38-49
• /
• 1996

In this paper, we propose a fast hierarchical image segmentation using mathematical morphology. The proposed segmentation method is composed of five basic steps; multi-thresholding, open-close by reconstructing, mode operation, marker extraction, and region decision. In the multi-thresholding, an input image is simplified by Lloyd clustering algorithm. The multi-thresholded image then is more simplified by open-close by reconstruction and mode operating. In the region decision, to which region each uncertainty pixel belongs finally is decided by a watershed algorithm. Experimental results show that the quality of the segmentation results by the proposed method is not inferior to that by the conventional method and the average times elapsed by the proposed method can be reduced by one tghird of those elapsed by the conventional method.