• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.1960-1963
• /
• 2003

For applications such as video surveillance and human computer interface, we propose an efficiently integrated method to detect and track faces. Various visual cues are combined to the algorithm: motion, skin color, global appearance and facial pattern detection. The ICA (Independent Component Analysis)-SVM (Support Vector Machine based pattern detection is performed on the candidate region extracted by motion, color and global appearance information. Simultaneous execution of detection and short-term tracking also increases the rate and accuracy of detection. Experimental results show that our detection rate is 91％ with very few false alarms running at about 4 frames per second for 640 by 480 pixel images on a Pentium IV 1㎓.

• Structural Engineering and Mechanics
• /
• v.60 no.4
• /
• pp.667-691
• /
• 2016

This study focuses on the damage detection of defect types in plate structures based on wavelet transform (WT) and curvelet transform (CT). In particular, for damage detection of structures these transforms have been developed since the last few years. In recent years, the CT approach has been also introduced in an attempt to overcome inherent limitations of traditional multi-scale representations such as wavelets. In this study, the performance of CT is compared with WT in order to demonstrate the capability of WT and CT in detection of defect types in plate structures. To achieve this purpose, the damage detection of defect types through defect shape in rectangular plate is investigated. By using the first mode shape of plate structure and the distribution of the coefficients of the transforms, the damage existence, the defect location and the approximate shape of defect are detected. Moreover, the accuracy and performance generality of the transforms are verified through using experimental modal data of a plate.

• International Journal of Aeronautical and Space Sciences
• /
• v.9 no.2
• /
• pp.114-120
• /
• 2008

In the vision-based recovery phase, a terminal guidance for the blended-wing UAV requires visual information of high accuracy. This paper presents the light source target design and detection algorithm for vision-based UAV recovery. We propose a recovery target design with red and green LEDs. This frame provides the relative position between the target and the UAV. The target detection algorithm includes HSV-based segmentation, morphology, and blob processing. These techniques are employed to give efficient detection results in day and night net recovery operations. The performance of the proposed target design and detection algorithm are evaluated through ground-based experiments.

• Journal of Korea Game Society
• /
• v.1 no.1
• /
• pp.42-48
• /
• 2001

Collision is a brief dynamic event consisting of the close approach of two or more objects or particles resulting in an abrupt change of momentum or exchange of energy because of interaction. Collisions play very important role in computer graphics, computer games and animations fields. Collisions can supply active interaction between cyberspace and real world and give much interests for making nice games so reasonable collision detection algorithms are needed. Collision detection algorithms should satisfy being fast and accuracy. In this paper, we survey the 2D collision detection algorithms between geometric models. We present several methods and system available for collision detection.

• ETRI Journal
• /
• v.39 no.6
• /
• pp.832-840
• /
• 2017

Recently, deep recurrent neural networks have achieved great success in various machine learning tasks, and have also been applied for sound event detection. The detection of temporally overlapping sound events in realistic environments is much more challenging than in monophonic detection problems. In this paper, we present an approach to improve the accuracy of polyphonic sound event detection in multichannel audio based on gated recurrent neural networks in combination with auditory spectral features. In the proposed method, human hearing perception-based spatial and spectral-domain noise-reduced harmonic features are extracted from multichannel audio and used as high-resolution spectral inputs to train gated recurrent neural networks. This provides a fast and stable convergence rate compared to long short-term memory recurrent neural networks. Our evaluation reveals that the proposed method outperforms the conventional approaches.

• Journal of IKEEE
• /
• v.18 no.2
• /
• pp.234-239
• /
• 2014

This paper proposes a new automatic detection method to inspect mura defects on display film surface using morphological image processing and labeling. This automatic detection method for mura defects on display films comprises 3 phases of preprocessing with morphological image processing, Gabor filtering, and labeling. Since distorted results could be obtained with the presence of non-uniform illumination, preprocessing step reduces illumination components using morphological image processing. In Gabor filtering, mura images are created with binary coded mura components using Gabor filters. Subsequently, labeling is a final phase of finding the mura defect area using the difference between large mura defects and values in the periphery. To evaluate the accuracy of the proposed detection method, detection rate was assessed by applying the method in 200 display film samples. As a result, the detection rate was high at about 95.5%. Moreover, the study was able to acquire reliable results using the Semu index for luminance mura in image quality inspection.

• Journal of Multimedia Information System
• /
• v.8 no.4
• /
• pp.211-220
• /
• 2021

The process of inspecting SMDs on the PCB boards improves the product quality, performance and reduces frequent issues in this field. However, undesirable scenarios such as assembly failure and device breakdown can occur sometime during the assembly process and result in costly losses and time-consuming. The detection of these components with a model based on deep learning may be effective to reduce some errors during the inspection in the manufacturing process. In this paper, YOLO models were used due to their high speed and good accuracy in classification and target detection. A SMD detection and classification method using YOLO networks based on robust data preprocessing and augmentation techniques to deal with various types of variation such as illumination and geometric changes is proposed. For 9 different components of data provided from a PCB manufacturer company, the experiment results show that YOLOv4 is better with fast detection and classification than YOLOv3.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.24 no.4
• /
• pp.392-400
• /
• 2021

Unlike optical equipment, SAR(Synthetic Aperture Radar) has the advantage of obtaining images in all weather, and object detection in SAR images is an important issue. Generally, deep learning-based object detection was mainly performed in real-valued network using only amplitude of SAR image. Since the SAR image is complex data consist of amplitude and phase data, a complex-valued network is required. In this paper, a complex-valued ResNet network is proposed. SAR image object detection was performed by combining the ROI transformer detector specialized for aerial image detection and the proposed complex-valued ResNet. It was confirmed that higher accuracy was obtained in complex-valued network than in existing real-valued network.

• International Journal of Computer Science & Network Security
• /
• v.22 no.6
• /
• pp.1-6
• /
• 2022

Scene change detection is an important topic because of the wide and growing range of its applications. Streaming services from many providers are increasing their capacity which causes the industry growth. The method for the scene change detection is described here and compared with the State-of-the-Art methods over the Open Video Scene Detection (OVSD) - an open dataset of Creative Commons licensed videos freely available for download and use to evaluate video scene detection algorithms. The proposed method is based on scene analysis using threshold values and smooth scene changes. A comparison of the presented method was conducted in this research. The obtained results demonstrated the high efficiency of the scene cut localization method proposed by authors, because its efficiency measured in terms of precision, recall, accuracy, and F-metrics score exceeds the best previously known results.

• Journal of Korea Multimedia Society
• /
• v.25 no.7
• /
• pp.903-911
• /
• 2022

A shrimp quality detection model using YOLOv4 deep learning algorithm is designed, which is superior in terms of network architecture, data processing and feature extraction. The shrimp images were taken and data expanded on their own, the LableImage platform was used for data annotation, and the network model was trained under the Darknet framework. Through comparison, the final performance of the model was all higher than other common target detection models, and its detection accuracy reached 93.7% with an average detection time of 47 ms, indicating that the method can effectively detect the quality of shrimp in the production process.