• Journal of Korea Society of Digital Industry and Information Management
• /
• v.19 no.3
• /
• pp.213-220
• /
• 2023

Face recognition is a very important process in video monitoring and is a type of biometric technology. It is mainly used for identification and security purposes, such as ID cards, licenses, and passports. The recognition process has many variables and is complex, so development has been slow. In this paper, we proposed a face recognition method using CNN, which has been re-examined due to the recent development of computers and algorithms, and compared with the feature comparison method, which is an existing face recognition algorithm, to verify performance. The proposed face search method is divided into a face region extraction step and a learning step. For learning, face images were standardized to 50×50 pixels, and learning was conducted while minimizing unnecessary nodes. In this paper, convolution and polling-based techniques, which are one of the deep learning technologies, were used for learning, and 1,000 face images were randomly selected from among 7,000 images of Caltech, and as a result of inspection, the final recognition rate was 98%.

• Economic and Environmental Geology
• /
• v.56 no.4
• /
• pp.397-408
• /
• 2023

A total of 64 specimens of trace fossils were collected from the Jinju Formation of the construction site of Jinju Aviation Industrial Complex, and from the Haman Formation of Namhae Gain-ri fossil site. The fossils are continuously and regularly meandering sine-curve in shape. The fossil varies in morphology: width between 0.2 and 5.6 mm, wavelength between 1.5 and 28 mm, and amplitude between 0.9 and 7.9 mm; the Jinju specimens are commonly wider than the Haman ones. The ratio of wavelength to amplitude is more or less regular regardless of width of the specimen, and the linear correlation of the ratios shows that the Jinju specimens fit better than the Haman specimens. Taking all morphometric parameters, specimens in all size ranges are temporarily identified as ichnospecies Cochlichnus anguineus. In order to obtain more distinct and clearer images of Cochlichnus, we selected two specimens and applied a new imaging technology RTI. For photography of the trace fossils, 50 to 80 images were taken per set with photometric lighting close to the surface and horizontally. RTI technology clearly showed that the images of tiny fossils were improved: the surface contrast become sharper and messy and unnecessary information disappeared. Currently, RTI technology is used in many fields including preservation of cultural properties and archaeology. As a consequence, we hope to apply this technique to the field of paleontology, especially to the study of trace fossils of very small size.

• Journal of Information Processing Systems
• /
• v.19 no.4
• /
• pp.427-438
• /
• 2023

Plenty of works have indicated that single image super-resolution (SISR) models relying on synthetic datasets are difficult to be applied to real scene text image super-resolution (STISR) for its more complex degradation. The up-to-date dataset for realistic STISR is called TextZoom, while the current methods trained on this dataset have not considered the effect of multi-scale features of text images. In this paper, a multi-scale and attention fusion model for realistic STISR is proposed. The multi-scale learning mechanism is introduced to acquire sophisticated feature representations of text images; The spatial and channel attentions are introduced to capture the local information and inter-channel interaction information of text images; At last, this paper designs a multi-scale residual attention module by skillfully fusing multi-scale learning and attention mechanisms. The experiments on TextZoom demonstrate that the model proposed increases scene text recognition's (ASTER) average recognition accuracy by 1.2% compared to text super-resolution network.

• Journal of KIISE:Computer Systems and Theory
• /
• v.32 no.1
• /
• pp.1-8
• /
• 2005

Image morphing to generate 2D transitions between images may be difficult even to express simple 3D transformations. In addition, previous view morphing method requires control points for postwarping, and is much affected by self- occlusion. This paper presents a new morphing algorithm that can generate automatically in-between scenes from un-calibrated images. Our algorithm rectifies input images based on the fundamental matrix, which is followed by linear interpolation with bilinear disparity map. In final, we generate in-between views by inverse mapping of homography between the rectified images. The proposed method nay be applied to photographs and drawings, because neither knowledge of 3D shape nor camera calibration, which is complex process generally, is required. The generated in-between views can be used in various application areas such as simulation system of virtual environment and image communication.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.6
• /
• pp.2480-2496
• /
• 2020

Considering that high-dose X-ray radiation during CT scans may bring potential risks to patients, in the medical imaging industry there has been increasing emphasis on low-dose CT. Due to complex statistical characteristics of noise found in low-dose CT images, many traditional methods are difficult to preserve structural details effectively while suppressing noise and artifacts. Inspired by the deep learning techniques, we propose a densely connected residual network (DCRN) for low-dose CT image noise cancelation, which combines the ideas of dense connection with residual learning. On one hand, dense connection maximizes information flow between layers in the network, which is beneficial to maintain structural details when denoising images. On the other hand, residual learning paired with batch normalization would allow for decreased training speed and better noise reduction performance in images. The experiments are performed on the 100 CT images selected from a public medical dataset-TCIA(The Cancer Imaging Archive). Compared with the other three competitive denoising algorithms, both subjective visual effect and objective evaluation indexes which include PSNR, RMSE, MAE and SSIM show that the proposed network can improve LDCT images quality more effectively while maintaining a low computational cost. In the objective evaluation indexes, the highest PSNR 33.67, RMSE 5.659, MAE 1.965 and SSIM 0.9434 are achieved by the proposed method. Especially for RMSE, compare with the best performing algorithm in the comparison algorithms, the proposed network increases it by 7 percentage points.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.4
• /
• pp.97-103
• /
• 2008

This study is to present a fainting motion recognizing method by using fish-eye lens images to sense emergency situations. The camera with fish-eye lens located at the center of the ceiling of the living room sends images, and then the foreground pixels are extracted by means of the adaptive background modeling method based on the Gaussian complex model, which is followed by tracing of outer points in the foreground pixel area and the elliptical mapping. During the elliptical tracing, the fish-eye lens images are converted to fluoroscope images. the size and location changes, and moving speed information are extracted to judge whether the movement, pause, and motion are similar to fainting motion. The results show that compared to using fish-eye lens image, extraction of the size and location changes. and moving speed by means of the conversed fluoroscope images has good recognition rates.

• Journal of Periodontal and Implant Science
• /
• v.47 no.1
• /
• pp.30-40
• /
• 2017

Purpose: Optical coherence tomography (OCT) is a noninvasive diagnostic technique that may be useful for both qualitative and quantitative analyses of the periodontium. Micro-computed tomography (micro-CT) is another noninvasive imaging technique capable of providing submicron spatial resolution. The purpose of this study was to present periodontal images obtained using ex vivo dental OCT and to compare OCT images with micro-CT images and histologic sections. Methods: Images of ex vivo canine periodontal structures were obtained using OCT. Biologic depth measurements made using OCT were compared to measurements made on histologic sections prepared from the same sites. Visual comparisons were made among OCT, micro-CT, and histologic sections to evaluate whether anatomical details were accurately revealed by OCT. Results: The periodontal tissue contour, gingival sulcus, and the presence of supragingival and subgingival calculus could be visualized using OCT. OCT was able to depict the surface topography of the dentogingival complex with higher resolution than micro-CT, but the imaging depth was typically limited to 1.2-1.5 mm. Biologic depth measurements made using OCT were a mean of 0.51 mm shallower than the histologic measurements. Conclusions: Dental OCT as used in this study was able to generate high-resolution, cross-sectional images of the superficial portions of periodontal structures. Improvements in imaging depth and the development of an intraoral sensor are likely to make OCT a useful technique for periodontal applications.

• Proceedings of the KAIS Fall Conference
• /
• 2006.11a
• /
• pp.220-224
• /
• 2006

The text to be included in the natural images has many important information in the natural image. Therefore, if we can extract the text in natural images, It can be applied to many important applications. In this paper, we propose a text region extraction method using pattern histogram of character-edge map. We extract the edges with the Canny edge detector and creates 16 kind of edge map from an extracted edges. And then we make a character-edge map of 8 kinds that have a character feature with a combination of an edge map. We extract text region using 8 kinds of character-edge map and 16 kind of edge map. Verification of text candidate region uses analysis of a character-edge map pattern histogram and structural feature of text region. The method to propose experimented with various kind of the natural images. The proposed approach extracted text region from a natural images to have been composed of a complex background, various letters, various text colors effectively.

• Journal of KIISE:Databases
• /
• v.29 no.2
• /
• pp.110-118
• /
• 2002

This paper presents a 3-D graphic database system based on XML that supports content-based retrievals of 3-D images, Most of graphics application systems are currently centered around the processing of 2-D images and research works on 3-D graphics are mainly concerned about the visualization aspects of 3-D image. They do not support the semantic modeling of 3-D objects and their spatial relations. In our data model, 3-D images are represented as compositions of 3-D graphic objects with associated spatial relations. Complex 3-D objects are mode]ed using a set of primitive 3-D objects rather than the lines and polygons that are found in traditional graphic systems. This model supports content-based retrievals of scenes containing a particular object or those satisfying certain spatial relations among the objects contained in them. 3-D images are stored in the database as XML documents using 3DGML DTD that are developed for modeling 3-D graphic data. Finally, this paper describes some examples of query executed in our Web-based prototype database system.

• Journal of Advanced Navigation Technology
• /
• v.26 no.2
• /
• pp.113-118
• /
• 2022

Motivated by people highlighting important phrases while reading or taking notes we propose a neural network training method by highlighting defective pixel areas to classify effectively defect types of images with complex background textures. To verify our proposed method we apply it to the problem of classifying the defect types of tire band fabric images that are too difficult to classify. In addition we propose a backlight highlighting technique which is tailored to the tire band fabric images. Backlight highlighting images can be generated by using both the GradCAM and simple image processing. In our experiment we demonstrated that the proposed highlighting method outperforms the traditional method in the view points of both classification accuracy and training speed. It achieved up to 13.4% accuracy improvement compared to the conventional method. We also showed that the backlight highlighting technique tailored for highlighting tire band fabric images is superior to a contour highlighting technique in terms of accuracy.