• Aerospace Engineering and Technology
• /
• v.6 no.1
• /
• pp.213-221
• /
• 2007

This paper showed how to analysis the object-based classification for wildfire fuel type map using Hyperion hyperspectral remote sensing data acquired in April, 2002 and compared the results of the object-based classification with the results of the pixel-based classification. Our methodological approach for wildfire fuel type map firstly processed correcting abnormal pixels and atypical bands and also calibrating atmospheric noise for enhanced image quality. Fuel type map is characterized by the results of the spectral mixture analysis(SMA). Object-based approach was based on segment-based endmember selection, while pixel-based method used standard SMA. To validate and compare, we used true-color high resolution orthoimagery.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.25 no.3
• /
• pp.461-469
• /
• 2001

Existing area-based stereo algorithms rely heavily on rectangular windows for computing correspondence. While the algorithms with the rectangular windows are efficient, they generate relatively large matching errors due to variations of disparity profiles near depth discontinuities and doesnt take into account local deformations of the windows due to projective distortion. In this paper, in order to deal with these problems, a new correlation function with 4 directional line masks, based on robust estimator, is proposed for the selection of potential matching points. These points is selected to consider depth discontinuities and reduce effects on outliers. The proposed matching method finds an arbitrarily-shaped variable window around a pixel in the 3d array which is constructed with the selected matching points. In addition, the method take into account the local deformation of the variable window with a constant disparity, and perform the estimation of sub-pixel disparities. Experiments with various synthetic images show that the proposed technique significantly reduces matching errors both in the vicinity of depth discontinuities and in continuously smooth areas, and also does not be affected drastically due to outlier and noise.

• Journal of Biosystems Engineering
• /
• v.39 no.3
• /
• pp.166-173
• /
• 2014

Purpose: In this study, we focused on the image processing method to determine the external quality of Fuji apples by identifying surface defects such as scabs and bruises. Method: A CCD camera was used to capture filter images with 24 different wavelengths ranging between 530 nm and 1050 nm. Image subtraction and division operations were performed to distinguish the defect area from the normal areas including calyx, stem, and glaring on the apple surface image. All threshold values of the image were examined to reveal the defect area of pretreated filter images. Results: The developed operation methods were [image (720 nm) - image (900 nm)]/image (700 nm) for bruise detection and [image (740 nm) - image (900 nm)]/image (590 nm) for scab detection, which revealed 81% and 90% recognition ratios, respectively. Conclusions: Our results showed several optimal wavelengths and image processing methods to detect Fuji apple surface defects such as bruises and scabs.

• Korean Journal of Remote Sensing
• /
• v.20 no.1
• /
• pp.47-55
• /
• 2004

We have constructed a level-1 processor to generate brightness temperatures using the direct-broadcast data from the passive microwave radiometer onboard Aqua satellite. Although 50-minute half-orbit data, called a granule, are being routinely produced by global data centers, to our knowledge, this is the first attempt to process 10-minute long direct-broadcast (DB) data. We found that the processor designed for a granule needs modification to apply to the DB data. The modification includes the correction to path number, the selection of land mask and the manipulation of dummy scans. Pixel-to-pixel comparison with a reference indicates the difference in brightness temperature of about 0.2 K rms and less than 0.05 K mean. The difference comes from the different length of data between 50-minute granule and about 10-minute DB data. In detail, due to the short data length, DB data do not always have correct cold sky mirror count. The DB processing system is automated to enable the near-real time generation of brightness temperatures within 5 minutes after downlink. Through this work, we would be able to enhance the use of AMSR-E data, thus serving the objective of direct-broadcast.

• Korea Science and Art Forum
• /
• v.37 no.3
• /
• pp.125-134
• /
• 2019

This study was conducted to organize a program to classify and analyze the characteristics of images for the automation of background music selection in the video content production process. The results and contents of the study are as follows: video characteristics are selected as subject category, emotion, pixel motion speed, color, and character material. Subject categories and feelings were extracted using Microsoft's Azure Video Indexer, Pixel Movement Speed was an Optional flow, Color was an Image Histogram for Image, and character materials was CNN(Convolutional Neural Network). The results of this study are significant in that video analysis was conducted to match background music in the recent content production process of 'Internet One-person Broadcasting Creators'.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.1
• /
• pp.42.2-43
• /
• 2021

We address a question of what are three main channels that can best translate other channels in ultraviolet (UV) and extreme UV (EUV) observations. For this, we compare the image translations among the nine channels of the Atmospheric Imaging Assembly on the Solar Dynamics Observatory using a deep learning model based on conditional generative adversarial networks. In this study, we develop 170 deep learning models: 72 models for single-channel input, 56 models for double-channel input, and 42 models for triple-channel input. All models have a single-channel output. Then we evaluate the model results by pixel-to-pixel correlation coefficients (CCs) within the solar disk. Major results from this study are as follows. First, the model with 131 Å shows the best performance (average CC = 0.84) among single-channel models. Second, the model with 131 and 1600 Å shows the best translation (average CC = 0.95) among double-channel models. Third, among the triple-channel models with the highest average CC (0.97), the model with 131, 1600, and 304 Å is suggested in that the minimum CC (0.96) is the highest. Interestingly they are representative coronal, photospheric, and chromospheric lines, respectively. Our results may be used as a secondary perspective in addition to primary scientific purposes in selecting a few channels of an UV/EUV imaging instrument for future solar satellite missions.

• Journal of Biosystems Engineering
• /
• v.17 no.1
• /
• pp.65-78
• /
• 1992

When using a computer vision system for a measurement, the geometrically distorted input image usually restricts the site and size of the measuring window. A geometrically distorted image caused by the image sensing and processing hardware degrades the accuracy of the visual measurement and prohibits the arbitrary selection of the measuring scope. Therefore, an image calibration is inevitable to improve the measuring accuracy. A calibration process is usually done via four steps such as measurement, modeling, parameter estimation, and compensation. In this paper, the efficient error calibration technique of a geometrically distorted input image was developed using a neural network. After calibrating a unit pixel, the distorted image was compensated by training CMLAN(Cerebellar Model Linear Associator Network) without modeling the behavior of any system element. The input/output training pairs for the network was obtained by processing the image of the devised sampled pattern. The generalization property of the network successfully compensates the distortion errors of the untrained arbitrary pixel points on the image space. The error convergence of the trained network with respect to the network control parameters were also presented. The compensated image through the network was then post processed using a simple DDA(Digital Differential Analyzer) to avoid the pixel disconnectivity. The compensation effect was verified using known sized geometric primitives. A way to extract directly a real scaled geometric quantity of the object from the 8-directional chain coding was also devised and coded. Since the developed calibration algorithm does not require any knowledge of modeling system elements and estimating parameters, it can be applied simply to any image processing system. Furthermore, it efficiently enhances the measurement accuracy and allows the arbitrary sizing and locating of the measuring window. The applied and developed algorithms were coded as a menu driven way using MS-C language Ver. 6.0, PC VISION PLUS library functions, and VGA graphic functions.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.12 no.12
• /
• pp.2337-2342
• /
• 2008

In this paper, we propose the novel approach to enhance the visual quality of the digital image with adaptively sharpening and removing the noise. Image enhancement is performed in two ways. The pixels in the high dynamics area are sharpened by the adaptive unsharp mask with the parameter, which is derived using the statistical information of the image. On the other hand, the proposed algorithm do not perform the sharpening process in the uniform area that may cause the undesired artifact due to noise amplification, rather it performs smoothing to suppress the noise in this area. The decision, which process will be applied at the pixel, is also controlled by the statistics of the pixel dynamics. The proposed algorithm enhances the visual quality almost automatically by sharpening and smoothing at the same time with less parameter selection.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 1998.06b
• /
• pp.92-97
• /
• 1998

For the efficient searching and browsing of digital video, it is essential to extract the internal structure of the video contents. As an example, a news video consists of several sections such as politics, economics, sports and others, and also each section consists of individual topics. With this information in hand, users can ore easily access the required video frames. This paper addresses the problem of automatic shot boundary detection and selection of representative frames (R-frames), which are the essential step in recognizing the internal structure of video contents. In the shot boundary detection, a new algorithm that have dual detectors which are designed specifically for the abrupt boundaries (cuts) and gradually changing bounaries respectively is proposed. Compared to the existing 미algorithms that mostly have tried to detect both types by a single mechanism, the proposed algorithm is proved to be more robust and accurate. Also in the problem of R-frame selection, simple mechanical approaches such as selecting one frame every other second have been adopted. However this approach often selects too many R-frames in static short, while drops important frames in dynamic shots. To improve the selection mechanism, a new R-frame selection algorithm that uses motion information extracted from pixel difference is proposed.