• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.12
• /
• pp.1919-1925
• /
• 2022

In this paper, we propose an image-to-image translation approach based on the conditional generative adversarial network for semantic segmentation. Semantic segmentation is the task of clustering parts of an image together which belong to the same object class. Unlike the traditional pixel-wise classification approach, the proposed method parses an input RGB image to its corresponding semantic segmentation mask using a pixel regression approach. The proposed method is based on the Pix2Pix image synthesis method. We employ residual connections-based convolutional neural network architectures for both the generator and discriminator architectures, as the residual connections speed up the training process and generate more accurate results. The proposed method has been trained and tested on the NYU-depthV2 dataset and could achieve a good mIOU value (49.5%). We also compare the proposed approach to the current methods in semantic segmentation showing that the proposed method outperforms most of those methods.

• International Journal of Control, Automation, and Systems
• /
• v.6 no.6
• /
• pp.828-835
• /
• 2008

We present a novel method for eye location by means of a two-level classifier scheme. Locating the eye by machine-inspection of an image or video is an important problem for Computer Vision and is of particular value to applications in biomedical imaging. Our method aims to overcome the significant challenge of an eye-location that is able to maintain high accuracy by disregarding highly variable changes in the environment. A first level of computational analysis processes this image context. This is followed by object detection by means of a two-class discrimination classifier(second algorithmic level).We have tested our eye location system using FERET and BioID database. We compare the performance of two-level classifier with that of non-level classifier, and found it's better performance.

• Journal of the Optical Society of Korea
• /
• v.12 no.3
• /
• pp.136-146
• /
• 2008

We proposed an adaptive steganography of an optical image using bit-planes and multichannel characteristics. The experiment's purpose was to compare the most popular methods used in optical steganography and to examine their advantages and disadvantages. In this paper we describe two digital methods: the first uses less significant bits(LSB) to encode hidden data, and in the other all blocks of $n{\times}n$ pixels are coded by using DCT(Digital Cosine Transformation), and two optical methods: double phase encoding and digital hologram watermarking with double binary phase encoding by using IFTA(Iterative Fourier Transform Algorithm) with phase quantization. Therefore, we investigated the complexity on bit plane and data, similarity insert information into bit planes. As a result, the proposed method increased the insertion capacity and improved the optical image quality as compared to fixing threshold and variable length method.

• Journal of Electrical Engineering and information Science
• /
• v.2 no.6
• /
• pp.146-155
• /
• 1997

In order to achieve scale- and rotation-invariance in recognizing unoccluded objects in binary images using Zernike moment features, an image of an object has often been normalized first by its zeroth-order moment (ZOM) or area. With elongated objects such as characters, a stroke width varies with the threshold value used, it becomes one or two pixels wider or thinner. The variations of the total area of the character becomes significant when the character is relatively thin with respect to its overall size, and the resulting normalized moment features are no longer reliable. This dilation/erosion effect is more severe when the object is not focused precisely. In this paper, we analyze the ZOM method and propose as a normalization method, the maximum enclosing circle (MEC) centered at the centroid of the character. We compare both the ZOM and MEC methods in their performance through various experiments.

• Journal of Electrical Engineering and information Science
• /
• v.2 no.6
• /
• pp.156-161
• /
• 1997

Zernike moments have been one of the most commonly used feature vectors for recognizing rotated patterns due to its rotation invariant characteristics. In order to reduce its expensive computational cost, several methods have been proposed to lower the complexity. One of the methods proposed by mukundan and K. R. Ramakrishnan[1], however, is not rotation invariant. In this paper, we propose another method that not only reduces the computational cost but preserves the rotation invariant characteristics. In the experiment, we compare our method with others, in terms of computing time and the accuracy of moment feature at different rotational angle of an object in image.

• Proceedings of the KIEE Conference
• /
• 1999.07g
• /
• pp.2941-2943
• /
• 1999

The application of the aerial images are to find the 3-D elevations. Image matching techniques such as Multi-resolution techniques, WCC (Weighted Cross-Correlation), NSSR (Narrow Search Sub-pixel Registration) that we know robustly apply to images which have enough features. But the method is not adaptive in images which have not enough features due to increasing of disparity errors. In this paper, we propose Disparity Interpolation that decrease disparity errors occurring in the area where images have not enough features. By using real aerial images we compare the result from existing image matching techniques to the result from proposed method.

• Journal of Korean Society of Transportation
• /
• v.12 no.1
• /
• pp.97-116
• /
• 1994

Today advanced traffic management systems are required because of a high increase in traffic demand. Accordingly, the objective of this study is to take advantage of image processing systems and present image processing methods available for collection of the data on traffic characteristics, and then to investigate the possibility of traffic flow analysis by means of comparison and analysis of measured traffic flow. Data were collected at two places of Daegu city and Kyongbu expressway by using VTR. Rear view (down stream) and frontal view (up stream) methods were employed to compare and analyze traffic characteristics including traffic volume, speed, time-headway, time-occupancy, and vehicle-length, by analysis of measured traffic flow and image processing respectively. Judging from the results obtained by this study, image processing techniques are sufficient for the analysis of traffic volume, but a frame grabber equipped with high speed processor is necessary as well, with low level system judged to be sufficient for traffic volume analysis.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.6
• /
• pp.980-985
• /
• 2002

In this paper, we analyzed satellite image with a high resolution compression performance. We need much time in a fast processing on vast satellite image pixel data. On compressing and decompressing, we should keep the information about road, building, forest, etc. In conclusion, we did analyze and compare the performance of compression and decompression for JPEG and WSQ(wavelet scalar quantization) method. As a result, we knew that WSQ was more efficient than JPEG.

• Proceedings of the KSRS Conference
• /
• 2002.10a
• /
• pp.559-564
• /
• 2002

This paper describes a new algorithm named as ray following algorithm which is applied for high-resolution satellite image simulation. The problems of the conventional ray tracing algorithm are pointed out especially when terrain elevations vary abruptly. The proposed algorithm follows the directional ray vector sequentially and thoroughly in order to determine the crossing point of the ray with the terrain surface. This way of sequential height comparison method is regarded as the only way to obtain accurate surface cross-section when a highly variant digital surface model is used. The experimental results show and compare the validities of the conventional and proposed algorithms.

• Journal of Convergence for Information Technology
• /
• v.11 no.12
• /
• pp.30-38
• /
• 2021

In this paper, we propose a method for extracting thermal data from thermal image and improving detection of heating equipment using the data. The main goal is to read the data in bytes from the thermal image file to extract the thermal data and the real image, and to apply the composite image obtained by synthesizing the image and data to the deep learning model to improve the detection accuracy of the heating facility. Data of KHNP was used for evaluation data, and Faster-RCNN is used as a learning model to compare and evaluate deep learning detection performance according to each data group. The proposed method improved on average by 0.17 compared to the existing method in average precision evaluation.As a result, this study attempted to combine national data-based thermal image data and deep learning detection to improve effective data utilization.