• Clinical Endoscopy
• /
• v.55 no.3
• /
• pp.390-400
• /
• 2022

Background/Aims: Previous artificial intelligence (AI) models attempting to segment gastric intestinal metaplasia (GIM) areas have failed to be deployed in real-time endoscopy due to their slow inference speeds. Here, we propose a new GIM segmentation AI model with inference speeds faster than 25 frames per second that maintains a high level of accuracy. Methods: Investigators from Chulalongkorn University obtained 802 histological-proven GIM images for AI model training. Four strategies were proposed to improve the model accuracy. First, transfer learning was employed to the public colon datasets. Second, an image preprocessing technique contrast-limited adaptive histogram equalization was employed to produce clearer GIM areas. Third, data augmentation was applied for a more robust model. Lastly, the bilateral segmentation network model was applied to segment GIM areas in real time. The results were analyzed using different validity values. Results: From the internal test, our AI model achieved an inference speed of 31.53 frames per second. GIM detection showed sensitivity, specificity, positive predictive, negative predictive, accuracy, and mean intersection over union in GIM segmentation values of 93%, 80%, 82%, 92%, 87%, and 57%, respectively. Conclusions: The bilateral segmentation network combined with transfer learning, contrast-limited adaptive histogram equalization, and data augmentation can provide high sensitivity and good accuracy for GIM detection and segmentation.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.13 no.4
• /
• pp.71-79
• /
• 2017

Recently, Fire watching and dangerous substances monitoring system has been being developed to enhance various fire related security. It is generally assumed that fire flame extraction plays a very important role on this monitoring system. In this study, we propose the fire flame extraction method of Non-Residential Facilities based on core object extraction in image. A core object is defined as a comparatively large object at center of the image. First of all, an input image and its decreased resolution image are segmented. Segmented regions are classified as the outer or the inner region. The outer region is adjacent to boundaries of the image and the rest is not. Then core object regions and core background regions are selected from the inner region and the outer region, respectively. Core object regions are the representative regions for the object and are selected by using the information about the region size and location. Each inner region is classified into foreground or background region by comparing its values of a color histogram intersection of the inner region against the core object region and the core background region. Finally, the extracted core object region is determined as fire flame object in the image. Through experiments, we find that to provide a basic measures can respond effectively and quickly to fire in non-residential facilities.

• Journal of Digital Contents Society
• /
• v.7 no.1
• /
• pp.9-14
• /
• 2006

In this paper, we propose the algorithm that automatically measures the height of the object to move on the base plane by using the geometric information. To extract a moving object from images, we use the difference image and morphology operation. The top and bottom point of an object are extracted by the histogram vertical projection in the extracted region. The two points, top and bottom, are used for measuring the height. Given the vanishing line of the ground plane, the vertical vanishing point, and at least one reference height in the scene; then the height of any point from the ground may be computed by specifying the image of the point and the image of the vertical intersection with the ground plane at that point. Through a confidence valuation of the height to be measured, we confirmed similar actual height and result in the simulation experiment.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.1
• /
• pp.117-124
• /
• 2008

We suggest CBIR(Content Based Image Retrieval) method using color and shape information. Using just one feature information may cause inaccuracy compared with using more than two feature information. Therefore many image retrieval system use many feature informations like color, shape and other features. We use two feature, HSI color information especially Hue value and CSS(Curvature Scale Space) as shape information. We search candidate image form DB which include feature information of many images. When we use two features, we could approach better result.

• The Journal of the Korea Contents Association
• /
• v.6 no.5
• /
• pp.20-28
• /
• 2006

This paper proposes visual-feature extraction for each band in wavelet domain with both spatial frequency features and multi resolution features, and the combination of visual features using fuzzy integral. In addition, it uses color feature expression method taking advantage of the frequency of the same color after color quantization for reducing quantization error, a disadvantage of the existing color histogram intersection method. Also, it is found that the final similarity can be represented in a linear combination of the respective factors(Homogram, color, energy) when each factor is independent one another. With respect to the combination patterns the fuzzy measurement is defined and the fuzzy integral is taken. Experiments are peformed on a database containing 1,000 color images. The proposed method gives better performance than the conventional method in both objective and subjective performance evaluation.

• The Journal of the Korea Contents Association
• /
• v.3 no.2
• /
• pp.24-30
• /
• 2003

Pre-existing authentication system such as token based method, knowledge-based and hybrid method have problems such as loss and wiretapping. for this, this paper describes the biometric authentication system which have the excellent convenience and security. In particular, a new biometric system by human face biometric features which have the non-enforcement and non-touch measurement is proposed. Firstly, facial features are extracted by Y- histogram and tilted face images we corrected by coordinate transformation and scaling has done for achieving independent of the camera positions. Secondly, feature vectors are extracted such as distance and intersection angles and similarities we measured by fuzzy relation matrix. finally, the effectiveness of this paper is demonstrated by experiments.

• Journal of the Korea Society of Computer and Information
• /
• v.7 no.3
• /
• pp.1-8
• /
• 2002

Due to the fast development in computer application technologies, a video is now being more widely used than ever in many areas. The current information analyzing systems are basically built to process text-based data. Thus, it has little bits Problems when it needs to correctly represent the ambiguity of a video, when it has to process a large amount of comments. or when it lacks the objectivity that the jobs require. We would like to purpose the method that is capable of analyze a large amount of video efficiently. To extract the color, we translate the color from RGB to HSI and use the information that matches with the representative colors. To extract the shape information, we use improved moment invariants(IMI) so that we can solve many problems of histogram intersection.

• Journal of Korea Multimedia Society
• /
• v.5 no.6
• /
• pp.616-624
• /
• 2002

An extraction method of central objects in the color images is proposed, in this paper. A central object is defined as a comparatively consist of the central object in the image. First of all. an input image and its decreased resolution images are segmented. Segmented regions are classified as the outer or the inner region. The outer region is adjacent regions are included by a same region in the decreased resolution image. Then core object regions and core background regions are selected from the inner region and the outer region respectively. Core object regions are the representative regions for the object and are selected by using the information about the information about the region size and location. Each inner regions is classified into foreground or background regions by comparing values of a color histogram intersection of the inner region against the core object region and the core background regions. The core object region and foreground regions consist of the central object in the image.

• Journal of Korea Multimedia Society
• /
• v.1 no.2
• /
• pp.173-182
• /
• 1998

As a result of remarkable developments in multimedia technology, the image database system that can efficiently retrieve image data becomes a core technology of information-oriented society. In this paper, we proposed the 2-phase Image Retrieval System considering both color and shape information as the method of image features extraction for content-based image data retrieval. At the first level, to get color information, with improving and extending the indexing method using color distribution characteristic suggested by Striker et al., i.e. the indexing method considering local color distribution characteristics, the system roughly classifies images through the improved method. At the second level, the system finally retrieves the most similar image from the image queried by the user using the shape information about the image groups classified at the first level. To extract the shape information, we use the Improved Moment Invariants (IMI) that manipulates only the pixels on the edges of objects in order to overcome two main problems of the existing Moment Invariant methods large amount of processing and rotation sensitiveness which can frequently be seen in the Directive Histogram Intersection technique suggested by Jain et al. Experiments have been conducted on 300 automobile images. And we could obtain the more improved results through the comparative test with other methods.

• Journal of the Korea Society of Computer and Information
• /
• v.5 no.2
• /
• pp.60-70
• /
• 2000

As a result of remarkable developments in computer technology, the image retrieval system that can efficiently retrieve image data becomes a core technology of information-oriented society. In this paper, we implemented the Hierarchical Image Retrieval System for content-based image data retrieval. At the first level, to get color information, with improving the indexing method using color distribution characteristic suggested by Striker et al., i.e. the indexing method considering local color distribution characteristics, the system roughly classifies images through the improved method. At the second level, the system finally retrieves the most similar image from the image queried by the user using the shape information about the image groups classified at the first level. To extract the shape information, we use the Improved Moment Invariants(IMI) that manipulates only the pixels on the edges of objects in order to overcome two main problems of the existing Moment Invariant methods large amount of processing and rotation sensitiveness which can frequently be seen in the Directive Histogram Intersection technique suggested by Jain et al. Experiments have been conducted on 300 automobile images And we could obtain the more improved results through the comparative test with other methods.