• Journal of applied mathematics & informatics
• /
• v.9 no.2
• /
• pp.793-806
• /
• 2002

Feature-based similarity retrieval become an important research issue in image database systems. The features of image data are useful to discrimination of images. In this paper, we propose the highspeed k-Nearest Neighbor search algorithm based on Self-Organizing Maps. Self-Organizing Maps (SOM) provides a mapping from high dimensional feature vectors onto a two-dimensional space. The mapping preserves the topology of the feature vectors. The map is called topological feature map. A topological feature map preserves the mutual relations (similarity) in feature spaces of input data. and clusters mutually similar feature vectors in a neighboring nodes. Each node of the topological feature map holds a node vector and similar images that is closest to each node vector. In topological feature map, there are empty nodes in which no image is classified. We experiment on the performance of our algorithm using color feature vectors extracted from images. Promising results have been obtained in experiments.

• Proceedings of the KIEE Conference
• /
• 2005.05a
• /
• pp.176-178
• /
• 2005

In the images of a video camera employed for surveillance, detecting targets by extracting foreground image is of great importance. The foreground regions detected, however, include not only moving targets but also their shadows. This paper presents a novel technique to detect shadow pixels in the foreground image of a video camera. The image characteristics of video cameras employed, a web-cam and a CCD, are first analysed in the HSV color space and a pixel-level shadow detection technique is proposed based on the analysis. Compared with existing techniques where unified criteria are used to all pixels, the proposed technique determines shadow pixels utilizing a fact that the effect of shadowing to each pixel is different depending on its brightness in background image. Such an approach can accommodate local features in an image and hold consistent performance even in changing environment. In experiments targeting pedestrians, the proposed technique showed better results compared with an existing technique.

• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.2
• /
• pp.145-148
• /
• 2011

In this paper, we proposed a search method for structured light pixels of omnidirectional structured light image. Since the omnidirectional structured light image is composed of several circular arc segments, the proposed algorithm searches the structured light pixels in radial direction rather than horizontal or vertical directions. The proposed search algorithm is based on the well-known Bresenham raster algorithm for line drawing in discrete integer space, thereby computation of the algorithm is very efficient. Comparison results between the proposed search algorithm and the conventional horizontal search are presented in experiments.

• Proceedings of the IEEK Conference
• /
• 2003.11a
• /
• pp.35-38
• /
• 2003

In this paper, I propose an automatic background extraction method and continuous background updating technique. Because there is a movement of a vehicle and a change of a background is feeble, the area moving through the time axis is looked for and a background and a vehicle image is divided. A way to give dynamically the threshold which divides the image frame into a vehicle image and the background in a space is enforced. Through the repetition of the above-mentioned process, the background pictorial image is gained. Using the karlman filter technique, the update is done so that a background image can obey a climate situation and an environmental change in day and night. A background image processed algorithm is better than the existent one. Through simulation, the feasibility of the algorithm has been verified.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.5
• /
• pp.327-334
• /
• 1991

This paper proposes an efficient method of implementing the generalized Hough transform (GHT), which has been hindered by an excessive computing load and a large memory requirement. The conventional algorithm requires a parameter space of 4 dimensions in detection a rotated, scaled, and translated object in an input image. Prior to the application of GHT to the input image, the proposed method determines the angle of rotation and the scaling factor of the test image using the proportion of the edge components between the reference image and test image. With the rotation angle and the scaling factor already determined, the parameter spaceis to be reduced to a simple array of 2 dimensions by applying the unit GHT only one time. The experiments with the image of airplanes reveal that both of the computing time and the requires memory size are reduced by 95 percent, without any degradatationof accuracy, compared with the conventional GHT algorithm.

• Journal of information and communication convergence engineering
• /
• v.13 no.1
• /
• pp.62-68
• /
• 2015

In this paper, we propose a watermarking method for protecting the ownership of three-dimensional (3D) content generated from depth and texture images. After selecting the target areas to preserve the watermark by depth-image-based rendering, the reference viewpoint image is moved right and left in the depth map until the maximum viewpoint change is obtained and the overlapped region is generated for marking space. The region is divided into four subparts and scanned. After applying discrete cosine transform, the watermarks are inserted. To extract the watermark, the viewpoint can be changed by referring to the viewpoint image and the corresponding depth image initially, before returning to the original viewpoint. The watermark embedding and extracting algorithm are based on quantization. The watermarked image is attacked by the methods of JPEG compression, blurring, sharpening, and salt-pepper noise.

• Proceedings of the IEEK Conference
• /
• 2000.07b
• /
• pp.894-897
• /
• 2000

In this paper, we present a content-based image retrieval method that is less sensitive to some rotations and translations of an image by using the fuzzy region segmentation. The algorithm retrieves similar images from a database using the two features of color and color spatial information. To index images, we use the average coordinates of color distribution to obtain the spatial information of each segmented region. Furthermore, we also propose the alternative to the ripple phenomenon, which is occurred in the conventional fuzzy region segmentation algorithm.

• Journal of the Semiconductor & Display Technology
• /
• v.22 no.1
• /
• pp.129-133
• /
• 2023

Image compression is a fundamental technique in the field of digital image processing, which will help to decrease the storage space and to transmit the files efficiently. Recently many deep learning techniques have been proposed to promise results on image compression field. Since many image compression techniques have artifact problems, this paper has compared two deep learning approaches to verify their performance experimentally to solve the problems. One of the approaches is a deep autoencoder technique, and another is a deep convolutional neural network (CNN). For those results in the performance of peak signal-to-noise and root mean square error, this paper shows that deep autoencoder method has more advantages than deep CNN approach.

• Journal of radiological science and technology
• /
• v.44 no.1
• /
• pp.31-37
• /
• 2021

This study aim of quantitative assessment of Noise Power Spectrum(NPS) and image characteristics of by acquired the optimal image for noise characteristics and quality assurance by using magnetic resonance imaging(MRI). MRI device was (MAGNETOM Vida 3.0T MRI; Siemense healthcare system; Germany) used and the head/neck shim MR receive coil were 20 channels coil and a diameter 200 mm hemisphere phantom. Frequency signal could be acquired the K-space trajectory image and white image for NPS. The T2 image highest quantitatively value for NPS finding of showed the best value of 0.026 based on the T2 frequency of 1.0 mm-1. The NPS acquired of showed that the T1 CE turbo image was 0.077, the T1 CE Conca2 turbo image was 0.056, T1 turbo image was 0.061, and the T1 Conca2 turbo image was 0.066. The assessment of NPS image characteristics of this study were to that could be used efficiently of the MRI and to present the quantitative evaluation methods and image noise characteristics of 3.0T MRI.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.1
• /
• pp.9-16
• /
• 2005

In this paper, we compare face recognition rate by distance change using Principal Component Analysis algorithm being input left and right image in stereo image. Change to YCbCr color space from RGB color space in proposed method and face region does detection. Also, after acquire distance using stereo image extracted face image's extension and reduce do extract robust face region, experimented recognition rate by using PCA algorithm. Could get face recognition rate of 98.61%(30cm), 98.91%(50cm), 99.05%(100cm), 99.90%(120cm), 97.31%(150cm) and 96.71%(200cm) by average recognition result of acquired face image. Therefore, method that is proposed through an experiment showed that can get high recognition rate if apply scale up or reduction according to distance.