• Korean Journal of Remote Sensing
• /
• v.28 no.4
• /
• pp.369-392
• /
• 2012

Recently, many target detection algorithms were developed for hyperspectral image. However, almost of these studies focused only accuracy from 1 or 2 data sets to validate and compare the algorithms although they give limited information to users. This study aimed to compare usability of target detection algorithms with various parameters. Five parameters were proposed to compare sensitivity in aspect of detection accuracy which are related with radiometric and spectral characteristics of target, background and image. Six target detection algorithms were compared in aspect of accuracy and efficiency (processing time) by variation of the parameters and image size, respectively. The results shown different usability of each algorithm by each parameter in aspect of accuracy. Second order statistics based algorithms needed relatively long processing time. Integrated usabilities of accuracy and efficiency were various by characteristics of target, background and image. Consequently, users would consider appropriate target detection algorithms by characteristics of data and purpose of detection.

• Journal of IKEEE
• /
• v.22 no.4
• /
• pp.1188-1194
• /
• 2018

This paper proposes an image enhancement-based license plate detection algorithm to improve the overall performance of system. Non-uniform illumination conditions have huge impact on overall plate detection system accuracy. In this paper, we propose an algorithm for color image enhancement-based license plate detection for improving accuracy of images degraded by excessively strong and low sunlight. Firstly, the image is enhanced by Multi-Scale Retinex algorithm. Secondly, a plate detection method is employed to take advantage of geometric properties of connected components, which can significantly reduce the undesired plate regions. Finally, intersection over union method is applied for detecting the accurate location of number plate. Experimental results show that the proposed method significantly improves the accuracy of plate detection system.

• Korean Journal of Remote Sensing
• /
• v.25 no.3
• /
• pp.271-285
• /
• 2009

Multi-temporal satellite imagery can be changed into a transform image that emphasizes the changed area only through the application of various change detection techniques. From the transform image, an automated change detection model calculates the optimal threshold value for classifying the changed and unchanged areas. However, the model can cause undesirable results when the histogram of the transform image is unbalanced. This is because the model uses a single threshold value in which the sign is either positive or negative and its value is constant (e.g. -1, 1), regardless of the imbalance between changed pixels. This paper proposes an advanced method that can improve accuracy by applying separate threshold values according to the increased or decreased range of the changed pixels. It applies multiple threshold values based on the cumulative producer's and user's accuracies in the automated binary change detection model, and the analyst can automatically extract more accurate optimal threshold values. Multi-temporal IKONOS satellite imagery for the Daejeon area was used to test the proposed method. A total of 16 transformation results were applied to the two study sites, and optimal threshold values were determined using accuracy assessment curves. The experiment showed that the accuracy of most transform images is improved by applying multiple threshold values. The proposed method is expected to be used in various study fields, such as detection of illegal urban building, detection of the damaged area in a disaster, etc.

• ETRI Journal
• /
• v.42 no.3
• /
• pp.411-419
• /
• 2020

In real-world intelligent transportation systems, accuracy in vehicle license plate detection and recognition is considered quite critical. Many algorithms have been proposed for still images, but their accuracy on actual videos is not satisfactory. This stems from several problematic conditions in videos, such as vehicle motion blur, variety in viewpoints, outliers, and the lack of publicly available video datasets. In this study, we focus on these challenges and propose a license plate detection and recognition scheme for videos based on a temporal matching prior network. Specifically, to improve the robustness of detection and recognition accuracy in the presence of motion blur and outliers, forward and bidirectional matching priors between consecutive frames are properly combined with layer structures specifically designed for plate detection. We also built our own video dataset for the deep training of the proposed network. During network training, we perform data augmentation based on image rotation to increase robustness regarding the various viewpoints in videos.

• ETRI Journal
• /
• v.41 no.5
• /
• pp.684-695
• /
• 2019

In a cloud environment, performance degradation, or even downtime, of virtual machines (VMs) usually appears gradually along with anomalous states of VMs. To better characterize the state of a VM, all possible performance metrics are collected. For such high-dimensional datasets, this article proposes a feature extraction algorithm based on unsupervised fuzzy linear discriminant analysis with kernel (UFKLDA). By introducing the kernel method, UFKLDA can not only effectively deal with non-Gaussian datasets but also implement nonlinear feature extraction. Two sets of experiments were undertaken. In discriminability experiments, this article introduces quantitative criteria to measure discriminability among all classes of samples. The results show that UFKLDA improves discriminability compared with other popular feature extraction algorithms. In detection accuracy experiments, this article computes accuracy measures of an anomaly detection algorithm (i.e., C-SVM) on the original performance metrics and extracted features. The results show that anomaly detection with features extracted by UFKLDA improves the accuracy of detection in terms of sensitivity and specificity.

• Smart Structures and Systems
• /
• v.28 no.6
• /
• pp.779-789
• /
• 2021

Cracks in railway sleeper are an inevitable condition and has a significant influence on the safety of railway system. Although the technology of railway sleeper condition monitoring using machine learning (ML) models has been widely applied, the crack recognition accuracy is still in need of improvement. In this paper, a two-stage method using edge detection and convolutional neural network (CNN) is proposed to reduce the burden of computing for detecting cracks in railway sleepers with high accuracy. In the first stage, the edge detection is carried out by using the 3×3 neighborhood range algorithm to find out the possible crack areas, and a series of mathematical morphology operations are further used to eliminate the influence of noise targets to the edge detection results. In the second stage, a CNN model is employed to classify the results of edge detection. Through the analysis of abundant images of sleepers with cracks, it is proved that the cracks detected by the neighborhood range algorithm are superior to those detected by Sobel and Canny algorithms, which can be classified by proposed CNN model with high accuracy.

• International Journal of Internet, Broadcasting and Communication
• /
• v.11 no.4
• /
• pp.76-85
• /
• 2019

We introduce a vision-based object detection method for real-time video surveillance system in low-end edge computing environments. Recently, the accuracy of object detection has been improved due to the performance of approaches based on deep learning algorithm such as Region Convolutional Neural Network(R-CNN) which has two stage for inferencing. On the other hand, one stage detection algorithms such as single-shot detection (SSD) and you only look once (YOLO) have been developed at the expense of some accuracy and can be used for real-time systems. However, high-performance hardware such as General-Purpose computing on Graphics Processing Unit(GPGPU) is required to still achieve excellent object detection performance and speed. To address hardware requirement that is burdensome to low-end edge computing environments, We propose sub-frame analysis method for the object detection. In specific, We divide a whole image frame into smaller ones then inference them on Convolutional Neural Network (CNN) based image detection network, which is much faster than conventional network designed forfull frame image. We reduced its computationalrequirementsignificantly without losing throughput and object detection accuracy with the proposed method.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.29 no.8
• /
• pp.640-647
• /
• 2018

This study analyzes the tracking accuracy(tracking errors) of fighter radar. Measurement error, detection failure, and radar cross section(RCS) fluctuation in radar measurements degrade the measurement quality and hence affect the tracking accuracy. Therefore, these radar measurement characteristics need to be considered when analyzing the tracking accuracy. In this paper, a method for analyzing the tracking accuracy is proposed; this method considers the detection error, detection probability, and RCS fluctuation. Results from experiments conducted with the proposed method show that the detection probability and RCS fluctuation affect tracking accuracy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.10
• /
• pp.5143-5158
• /
• 2018

Improving the intrusion detection system (IDS) is a pressing need for cyber security world. With the growth of computer networks, there are constantly daily new attacks. Machine Learning (ML) is one of the most important fields which have great contribution to address the intrusion detection issues. One of these issues relates to the imbalance of the diverse classes of network traffic. Accuracy paradox is a result of training ML algorithm with imbalanced classes. Most of the previous efforts concern improving the overall accuracy of these models which is truly important. However, even they improved the total accuracy of the system; it fell in the accuracy paradox. The seriousness of the threat caused by the minor classes and the pitfalls of the previous efforts to address this issue is the motive for this work. In this paper, we consolidated stratified sampling, cost function and weighted Support Vector Machine (WSVM) method to address the accuracy paradox of ID problem. This model achieved good results of total accuracy and superior results in the small classes like the User-To-Remote and Remote-To-Local attacks using the improved version of the benchmark dataset KDDCup99 which is called NSL-KDD.

• Journal of Korea Multimedia Society
• /
• v.24 no.7
• /
• pp.890-902
• /
• 2021

Although the object detection accuracy with still images has been significantly improved with the advance of deep learning techniques, the object detection problem with video data remains as a challenging problem due to the real-time requirement and accuracy drop with occlusion. In this research, we propose a method in pig detection for video monitoring environment. First, we determine a motion, from a video data obtained from a tilted-down-view camera, based on the average size of each pig at each location with the training data, and extract key frames based on the motion information. For each key frame, we then apply YOLO, which is known to have a superior trade-off between accuracy and execution speed among many deep learning-based object detectors, in order to get pig's bounding boxes. Finally, we merge the bounding boxes between consecutive key frames in order to reduce false positive and negative cases. Based on the experiment results with a video data set obtained from a pig farm, we confirmed that the pigs could be detected with an accuracy of 97% at a processing speed of 37fps.