• Journal of the Korea Society of Computer and Information
• /
• v.21 no.6
• /
• pp.21-28
• /
• 2016

Circle detection has long been studied as one of fundamental image processing applications. It is used in divers areas including industrial inspection, medial image analysis, radio astronomy data analysis, and other object recognition applications. The most widely used class of circle detection techniques is the circle Hough transform and its variants. Management of 3 dimensional parameter histogram used in these methods brings about spatial and temporal overheads, and a lot of studies have dealt the problem. This paper proposes a robust circle detection method using maximal symmetry property of circle. The basic idea is that if perpendicular bisectors of pairs of edges are accumulated in image space, center of circle is determined to be the location of highest accumulation. However, directly implementing the idea in image space requires a lot of calculations. The method of this paper reduces the number of calculations by mapping the perpendicular bisectors into parameter space, selecting small number of parameters, and mapping them inversely into image space. Test on 22 images shows the calculations of the proposed method is 0.056% calculations of the basic idea. The test images include simple circles, multiple circles with various sizes, concentric circles, and partially occluded circles. The proposed method detected circles in various situations successfully.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.747-761
• /
• 2021

With the increasing application of UAVs in intelligent transportation systems, vehicle detection for aerial images has become an essential engineering technology and has academic research significance. In this paper, a vehicle detection method for aerial images based on the YOLOv4 deep learning algorithm is presented. At present, the most known datasets are VOC (The PASCAL Visual Object Classes Challenge), ImageNet, and COCO (Microsoft Common Objects in Context), which comply with the vehicle detection from UAV. An integrated dataset not only reflects its quantity and photo quality but also its diversity which affects the detection accuracy. The method integrates three public aerial image datasets VAID, UAVD, DOTA suitable for YOLOv4. The training model presents good test results especially for small objects, rotating objects, as well as compact and dense objects, and meets the real-time detection requirements. For future work, we will integrate one more aerial image dataset acquired by our lab to increase the number and diversity of training samples, at the same time, while meeting the real-time requirements.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.8
• /
• pp.2199-2213
• /
• 2024

To enhance the quality of defect detection for Printed Circuit Board Assembly (PCBA) during electronic product manufacturing, this study primarily focuses on optimizing the YOLOv7-based method for PCBA defect detection. In this method, the Mish, a smoother function, replaces the Leaky ReLU activation function of YOLOv7, effectively expanding the network's information processing capabilities. Concurrently, a Squeeze-and-Excitation attention mechanism (SEAM) has been integrated into the head of the model, significantly augmenting the precision of small target defect detection. Additionally, considering angular loss, compared to the CIoU loss function in YOLOv7, the SIoU loss function in the paper enhances robustness and training speed and optimizes inference accuracy. In terms of data preprocessing, this study has devised a brightness adjustment data enhancement technique based on split-filtering to enrich the dataset while minimizing the impact of noise and lighting on images. The experimental results under identical training conditions demonstrate that our model exhibits a 9.9% increase in mAP value and an FPS increase to 164 compared to the YOLOv7. These indicate that the method proposed has a superior performance in PCBA defect detection and has a specific application value.

• The Bulletin of The Korean Astronomical Society
• /
• v.41 no.1
• /
• pp.75.4-76
• /
• 2016

DEEP-South (Deep Ecliptic patrol of the Southern sky) is one of the secondary science projects of KMTNet (Korea Microlensing Telescope Network). The objective of this project is twofold, the physical characterization and the discovery of small Solar System bodies, focused on NEOs (Near Earth objects). In order to achieve the goals, we are implementing a software package to detect and report moving objects in the $18k{\times}18k$ mosaic CCD images of KMTNet. In this paper, we present preliminary results of the moving object detection experiments using the prototype MODP (Moving Object Detection Program). We utilize multiple images that are being taken at three KMTNet sites, towards the same target fields (TFs) obtained at different epochs. This prototype package employs existing softwares such as SExtractor (Source-Extracto) and SCAMP (Software for Calibrating Astrometry and Photometry); SExtractor generates catalogs, while SCAMP conducts precision astrometric calibration, then MODP determines if a point source is moving. We evaluated the astrometric accuracy and efficiency of the current version of MODP. The plan for upgrading MODP will also be mentioned.

• Journal of Korea Multimedia Society
• /
• v.8 no.8
• /
• pp.1057-1068
• /
• 2005

This paper proposes a real-time face detection system which detects multiple faces from low resolution video such as web-camera video. First, It finds face region candidates by using AdaBoost based object detection method which selects a small number of critical features from a larger set. Next, it generates reduced feature vector for each face region candidate by using principle component analysis. Finally, it classifies if the candidate is a face or non-face by using SVM(Support Vector Machine) based binary classification. According to experiment results, the proposed method achieves real-time face detection from low resolution video. Also, it reduces the false detection rate than existing methods by using PCA and SVM based face classification step.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.37 no.5
• /
• pp.389-395
• /
• 2019

Unlike aerial images covering small region, satellite data show high potential to detect urban scale geospatial changes. The change detection using satellite images can be carried out using single image or stereo images. The single image approach is based on radiometric differences between two images of different times. It has limitations to detect building level changes when the significant occlusion and relief displacement appear in the images. In contrast, stereo satellite data can be used to generate DSM (Digital Surface Model) that contain information of relief-corrected objects. Therefore, they have high potential for the object change detection. Therefore, we carried out a study for the change detection over an urban area using stereo satellite data of two different times. First, the RPC correction was performed for two DSMs generation via stereo image matching. Then, D-DSM (Differential DSM) was generated by differentiating two DSMs. The D-DSM was used for the topographic change detection and the performance was checked by applying different height thresholds to D-DSM.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.29 no.2
• /
• pp.139-146
• /
• 2023

Maritime object detection systems, which detects small maritime obstacles such as fish farm buoys and visualizes distance and direction, is equipped with a 3-axis gimbal to compensate for errors caused by hull motion, but there is a limit to distance error corrections necessitated by the vertical movement of the camera and the maritime object due to wave motions. Therefore, in this study, the distance error of maritime object detection systems caused by the movement of the water surface according to the external environment is analyzed and corrected using average filter and moving average filter. Random numbers following a Gaussian standard normal distribution were added to or subtracted from the image coordinates to reproduce the rise or fall of the buoy under irregular waves. The distance calculated according to the change of image coordinates, the predicted distance through the average filter and the moving average filter, and the actual distance measured by laser distance meter were compared. In phases 1 and 2, the error rate increased to a maximum of 98.5% due to the changes of image coordinates due to irregular waves, but the error rate decreased to 16.3% with the moving average filter. This error correction capability was better than with the average filter, but there was a limit due to failure to respond to the distance change. Therefore, it is considered that use of the moving average filter to correct the distance error of the maritime object detection system will enhance responses to the real-time distance change and greatly improve the error rate.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.13 no.4
• /
• pp.315-321
• /
• 2009

In this paper, a level set based energy functional is proposed, the minimization of which results in simultaneous reference background image modeling and foreground segmentation. Due to the mutual constraint of the two processes, a good estimate of the background can be obtained with a small number of frames, and due to the use of the level set, an Euler-Lagrange equation that directly solves the problem can be derived.

• Journal of radiological science and technology
• /
• v.46 no.4
• /
• pp.277-285
• /
• 2023

In this study, pneumonia identification networks with the small number of layers were constructed by using chest X-ray images. The networks had similar trainable-parameters, and the performance of the trained models was quantitatively evaluated with the modification of the network architectures. A total of 6 networks were constructed: convolutional neural network (CNN), VGGNet, GoogleNet, residual network with identity blocks, ResNet with bottleneck blocks and ResNet with identity and bottleneck blocks. Trainable parameters for the 6 networks were set in a range of 273,921-294,817 by adjusting the output channels of convolution layers. The network training was implemented with binary cross entropy (BCE) loss function, sigmoid activation function, adaptive moment estimation (Adam) optimizer and 100 epochs. The performance of the trained models was evaluated in terms of training time, accuracy, precision, recall, specificity and F1-score. The results showed that the trained models with the small number of layers precisely detect pneumonia from chest X-ray images. In particular, the overall quantitative performance of the trained models based on the ResNets was above 0.9, and the performance levels were similar or superior to those based on the CNN, VGGNet and GoogleNet. Also, the residual blocks affected the performance of the trained models based on the ResNets. Therefore, in this study, we demonstrated that the object detection networks with the small number of layers are suitable for detecting pneumonia using chest X-ray images. And, the trained models based on the ResNets can be optimized by applying appropriate residual-blocks.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1141-1147
• /
• 2020

The hornet species are so similar in shape that they are difficult for non-experts to classify, and because the size of the objects is small and move fast, it is more difficult to detect and classify the species in real time. In this paper, we developed a system that classifies hornets species in real time based on a deep learning algorithm using a boundary box. In order to minimize the background area included in the bounding box when labeling the training image, we propose a method of selecting only the head and body of the hornet. It also experimentally compares existing boundary box-based object recognition algorithms to find the best algorithms that can detect wasps in real time and classify their species. As a result of the experiment, when the mish function was applied as the activation function of the convolution layer and the hornet images were tested using the YOLOv4 model with the Spatial Attention Module (SAM) applied before the object detection block, the average precision was 97.89% and the average recall was 98.69%.