• Journal of KIISE:Software and Applications
• /
• v.29 no.3
• /
• pp.145-155
• /
• 2002

In content-based image retrieval (CBIR), most conventional approaches assume a linear relationship between different features and require users themselves to assign the appropriate weights to each feature. However, the linear relationship assumed between the features is too restricted to accurately represent high-level concepts and the intricacies of human perception. In this paper, a neural network-based image retrieval (NNIR) model is proposed. It has been developed based on a human-computer interaction approach to CBIR using a radial basis function network (RBFN). By using the RBFN, this approach determines the nonlinear relationship between features and it allows the user to select an initial query image and search incrementally the target images via relevance feedback so that more accurate similarity comparison between images can be supported. The experiment was performed to calculate the level of recall and precision based on a database that contains 1,015 images and consists of 145 classes. The experimental results showed that the recall and level of the proposed approach were 93.45% and 80.61% respectively, which is superior than precision the existing approaches such as the linearly combining approach, the rank-based method, and the backpropagation algorithm-based method.

• Journal of Korean Society for Geospatial Information Science
• /
• v.14 no.4 s.38
• /
• pp.27-36
• /
• 2006

Conventionally, in order to get accurate geolocation of satellite images we need a set of ground control points with respect to individual scenes. In this paper, we tested the possibilities of modeling satellite orbits from individual scenes by establishing a sensor model for one scene and by applying the model, which was derived from the same orbital segment, to other scenes that has been acquired from the same orbital segment. We investigated orbit-attitude models with several interpolation methods and with various parameter sets to be adjusted. We used 7 satellite images of SPOT-3 with a length of 420km and ground control points acquired from GPS surveying. Results of the conventional individual scene modeling hardly introduced differences among different interpolation methods and different adjustment parameter sets. As the results of orbit modeling, the best model was the one with Lagrange interpolation for position/velocity and linear interpolation for attitude and with position/angle bias as parameter sets. The best model showed that it is possible to model orbital segments of 420km with ground control points measured within one scene (60km).

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.14 no.11
• /
• pp.5834-5840
• /
• 2013

Along with the development of high-speed, wired and wireless Internet technology, various harmful images in a form of photos and video clips have become prevalent these days. In this paper, we suggest a method of automatically detecting adult images by extracting woman's nipple areas which represent obscenity of the image. The suggested algorithm first segments skin color areas in the $YC_bC_r$ color space from input images and extracts nipple's candidate areas from the segmented skin areas through the suggested nipple map. We then select real nipple areas by using geometrical information and determines input images as harmful images if they contain nipples. Experimental results show that the suggested nipple map-based method effectively detects adult images.

• Journal of Internet Computing and Services
• /
• v.19 no.3
• /
• pp.35-42
• /
• 2018

Deep learning is known as a method of high accuracy among several methods for image classification. In this paper, we propose a method of enhancing the accuracy of image classification using CNN with a batch normalization method for classification of images using deep CNN (Convolutional Neural Network). In this paper, we propose a method to add a batch normalization layer to existing neural networks to enhance the accuracy of image classification. Batch normalization is a method to calculate and move the average and variance of each batch for reducing the deflection in each layer. In order to prove the superiority of the proposed method, Accuracy and mAP are measured by image classification experiments using five image data sets SHREC13, MNIST, SVHN, CIFAR-10, and CIFAR-100. Experimental results showed that the CNN with batch normalization is better classification accuracy and mAP rather than using the conventional CNN.

• Journal of Digital Contents Society
• /
• v.16 no.1
• /
• pp.127-135
• /
• 2015

The development of stereoscopic display hardwares and 3D authoring softwares expands its application areas from particular virtual simulation applications to general movies, games, advertising applications. However, the binocular-based 3D stereoscopic images cause fatigue to viewers. Recent performed many research results about the binocular stereoscopy's depth perception and viewers' fatigue are derived from experimental users studies. In some results, watching and making guidelines for 3D stereoscopic imaging contents are introduced. The 3D stereoscopic-related contents have the contradictory aspects, which are audiences' pursuit of a tolerable minimum fatigue and producer's its of excessive depth changes for providing viewers' immersion. This paper provides user experiments and analysis data in aspects of 3D depth changes. For use of producers, a safety zone and translational velocity of 3D depth changes are introduced. Also, on the viewer side, we present the depth change adaptation time by using an EEG device.

• Journal of the Korean Chemical Society
• /
• v.54 no.4
• /
• pp.465-470
• /
• 2010

Using interesting video clips corresponding to lesson subjects for students who favour visual representation is one of the good methods to enhance students' preference for science class. There are many moving picture web sites to get video clips easily via internet and 'YouTube' is very popular and one of the largest reservoir. In this study, every student in the 'Chemistry I' class, which is a class for 11th grade, was requested to search a video clip corresponding to lesson subjects and to make a presentation in the class. After 1st semester, students' response about the class using YouTube was examined by survey. As a result, students preferred and were interested in the class using YouTube than class centered on textbook. And students preferred YouTube clips showing unusual experiments that were related with contents of subject. In addition, experiments and watching their real phenomena were an interesting factor and helpful factor of learning chemistry in YouTube video clips, respectively. However, translation of English used in the video clips seemed to be a difficult part for students.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2008.08a
• /
• pp.73-76
• /
• 2008

We present a area-based method for stereo matching using varying weights. A central problem in a area-based stereo matching is different result from selecting a window size. Most of the previous window-based methods iteratively update windows. However, the iterative methods very sensitive the initial disparity estimation and are computationally expensive. To resolve this problem, we proposed a new function to assign weights to pixels using features. To begin with, we extract features in a given stereo images based on edge. We adjust the weights of the pixels in a given window based on correlation of the stereo images. Then, we match pixels in a given window between the reference and target images of a stereo pair. The proposed method is compared to existing matching strategies using both synthetic and real images. The experimental results show the improved accuracy of the proposed method.

• Journal of Broadcast Engineering
• /
• v.15 no.1
• /
• pp.76-88
• /
• 2010

Recently, LED background image devices are used for the production of broadcasting programs. But the effective advantages of LED background image devices are not realized for good HD video quality because the correlations between lighting sources and LED background image devices are not analyzed yet. In this paper, we analyze the color change and light intensity that lighting sources and the light intensity ratio of LED background images influence on a subject, making use of equipments used in real broadcasting, and examine the correlations between HD video quality and the light intensity ratio of LED background images. As a result in this experiment, the video is most expressed having an 20[IRE] effect on a subject when the light intensity of LED background image is 6%. When the LED background image is light, the light and darkness of video is distorted and when the LED background image is dark, the colors of image are distorted.

• Korean Journal of Remote Sensing
• /
• v.15 no.3
• /
• pp.183-194
• /
• 1999

The Brightness, color and texture included in a multispectral satellite data are used as important factors to analyze and to apply the image data for a proper use. One of the most significant process in the satellite data analysis using texture or color information is to extract features effectively expressing the information of original image. It was described in this paper that six features were introduced to extract useful features from the analysis of the satellite data, and also a classification network using the back-propagation neural network was constructed to evaluate the classification ability of each vector feature in SPOT imagery. The vector features were adopted from the training set selection for the interesting region, and applied to the classification process. The classification results showed that each vector feature contained many merits and demerits depending on each vector's characteristics, and each vector had compatible classification ability. Therefore, it is expected that the color and texture features are effectively used not only in the classification process of satellite imagery, but in various image classification and application fields.

• Korean Journal of Remote Sensing
• /
• v.35 no.5_1
• /
• pp.727-736
• /
• 2019

In this paper, we present how to create a point cloud for a target area using multiple unmanned aerial vehicle images and to remove the gaps and overlapping points between datasets. For this purpose, first, IBA (Incremental Bundle Adjustment) technique was applied to correct the position and attitude of UAV platform. We generate a point cloud by using MDR (Multi-Dimensional Relaxation) matching technique. Next, we register point clouds based on observation vectors between stereo images by doing this we remove gaps between point clouds which are generated from different stereo pairs. Finally, we applied an occupancy grids based integration algorithm to remove duplicated points to create an integrated point cloud. The experiments were performed using UAV images, and our experiments show that it is possible to remove gaps and duplicate points between point clouds generated from different stereo pairs.