• Journal of the Korea Society of Computer and Information
• /
• v.26 no.3
• /
• pp.127-135
• /
• 2021

Raising large numbers of animals in very narrow environments such as laying hens house can be very damaged from small environmental change. Previously researched about laying hens sound analysis system has a problem for applying to the laying hens house because considering only the limited situation of laying hens house. In this paper, to solve the problem, we propose a new laying hens sound analysis model using MFCC feature vector. This model can detect 7 situations that occur in actual laying hens house through 9 kinds of laying hens sound analysis. As a result of the performance evaluation of the proposed laying hens sound analysis model, the average AUC was 0.93, which is about 43% higher than that of the frequency feature analysis method.

• Journal of Internet Computing and Services
• /
• v.24 no.1
• /
• pp.49-59
• /
• 2023

Recently, attention to the pandemic situation represented by COVID-19 emerged problems caused by unexpected shortage of medical personnel. In this paper, we present a method for diagnosing the presence or absence of lesional sign on PA chest X-ray images as computer vision solution to support diagnosis tasks. Method for visual anomaly detection based on feature modeling can be also applied to X-ray images. With extracting feature vectors from PA chest X-ray images and divide to patch unit, region-specific abnormality can be detected. As preliminary experiment, we created simulation data set containing multiple objects and present results of the comparative experiments in this paper. We present method to improve both efficiency and performance of the process through hard masking of patch features to aligned images. By summing up regional specificity and global anomaly detection results, it shows improved performance by 0.069 AUROC compared to previous studies. By aggregating region-specific and global anomaly detection results, it shows improved performance by 0.069 AUROC compared to our last study.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.3
• /
• pp.251-256
• /
• 2019

In this paper we proposed system and algorithm to recognize hand gestures based on the millimeter wave that is in 65GHz bandwidth. The proposed system is composed of millimeter wave radar board, analog to data conversion and data capture board and notebook to perform gesture recognition algorithms. As feature vectors in proposed algorithm. we used global and local zernike moment descriptor which are robust to distort by rotation of scaling of 2D data. As Experimental result, performance of the proposed algorithm is evaluated and compared with those of algorithms using single global or local zernike descriptor as feature vectors. In analysis of confusion matrix of algorithms, the proposed algorithm shows the better performance in comparison of precision, accuracy and sensitivity, subsequently total performance index of our method is 95.6% comparing with another two mehods in 88.4% and 84%.

• Journal of Biomedical Engineering Research
• /
• v.23 no.1
• /
• pp.39-47
• /
• 2002

The objective of this study is to provide a Precise form of spatial normalization with affine transformation. The quantitative comparison of the brain architecture across different subjects requires a common coordinate system. For the common coordinate system, not only global brain but also a local region of interest should be spatially normalized. Registration using mutual information generally matches the whose brain well. However. a region of interest may not be normalized compared to the feature-based methods with the landmarks. The hybrid method of this Paper utilizes feature information of the local region as well as intensity similarity. Central gray nuclei of a brain including copus callosum, which is used for feature in Schizophrenia detection, is appropriately normalized by the hybrid method. In the results section. our method is compared with mutual information only method and Talairach mapping with schizophrenia Patients. and is shown how it accurately normalizes feature .

• Journal of Convergence for Information Technology
• /
• v.12 no.3
• /
• pp.17-24
• /
• 2022

Mobile malicious apps are increasing rapidly, and Android, which accounts for most of the global mobile OS market, is becoming a major target of mobile cyber security threats. Therefore, in order to cope with rapidly evolving malicious apps, there is a need for detection techniques of malicious apps using machine learning, one of artificial intelligence implementation technologies. In this paper, we propose a selected feature method using feature selection and feature extraction that can improve the detection performance of malicious apps. In the feature selection process, the detection performance improved according to the number of features, and the API showed relatively better detection performance than the permission. Also combining the two characteristics showed high precision of over 93% on average, confirming that the appropriate combination of characteristics could improve the detection performance.

• International Journal of Automotive Technology
• /
• v.8 no.4
• /
• pp.493-504
• /
• 2007

This paper proposes a stereo vision-based forward obstacle detection and distance measurement method. In general, stereo vision-based obstacle detection methods in automotive applications can be classified into two categories: IPM (Inverse Perspective Mapping)-based and disparity histogram-based. The existing disparity histogram-based method was developed for stop-and-go applications. The proposed method extends the scope of the disparity histogram-based method to highway applications by 1) replacing the fixed rectangular ROI (Region Of Interest) with the traveling lane-based ROI, and 2) replacing the peak detection with a constant threshold with peak detection using the threshold-line and peakness evaluation. In order to increase the true positive rate while decreasing the false positive rate, multiple candidate peaks were generated and then verified by the edge feature correlation method. By testing the proposed method with images captured on the highway, it was shown that the proposed method was able to overcome problems in previous implementations while being applied successfully to highway collision warning/avoidance conditions, In addition, comparisons with laser radar showed that vision sensors with a wider FOV (Field Of View) provided faster responses to cutting-in vehicles. Finally, we integrated the proposed method into a longitudinal collision avoidance system. Experimental results showed that activated braking by risk assessment using the state of the ego-vehicle and measuring the distance to upcoming obstacles could successfully prevent collisions.

• Proceedings of the KSRS Conference
• /
• 2007.10a
• /
• pp.286-289
• /
• 2007

This study is to develop a cloud detection algorit1un for COMS and it is currently tested by using MODIS level 2B and MTSAT-1R satellite radiance data. Unlike many existing cloud detection schemes which use a threshold method and traditional statistical methods, in this study a feed-forward neural network method with back-propagation algorit1un is used. MODIS level 2B products are matched with feature information of five-band MTSAT 1R image data to form the training dataset. The neural network is trained over the global region for the period of January to December in 2006 with 5 km spatial resolution. The main results show that this model is capable to detect complex cloud phenomena. And when it is applied to seasonal images, it shows reliable results to reflect seasonal characteristics except for snow cover of winter. The cloud detection by the neural network method shows 90% accuracy compared to the MODIS products.

• Journal of Broadcast Engineering
• /
• v.5 no.1
• /
• pp.31-40
• /
• 2000

In this paper, we propose an edge histogram to efficiently represent the edge distribution in the image for MPEG-7. To this end, we adopt global, semi-global, and local edge histogram bins. Also, we extract the edge information from the image in terms of image blocks rather than pixels, which reduces the extraction complexity and is also applicable to the block-based compression standards such as MPEG-1, and 2. Experimental results show that the proposed method yields better retrieval accuracy and feature extraction speed comparing to other non-homogeneous texture descriptors of MPEG-7 including the wavelet-based descriptor and local edge-based descriptor.

• ETRI Journal
• /
• v.36 no.6
• /
• pp.913-923
• /
• 2014

This paper proposes a global mapping algorithm for multiple robots from an omnidirectional-vision simultaneous localization and mapping (SLAM) approach based on an object extraction method using Lucas-Kanade optical flow motion detection and images obtained through fisheye lenses mounted on robots. The multi-robot mapping algorithm draws a global map by using map data obtained from all of the individual robots. Global mapping takes a long time to process because it exchanges map data from individual robots while searching all areas. An omnidirectional image sensor has many advantages for object detection and mapping because it can measure all information around a robot simultaneously. The process calculations of the correction algorithm are improved over existing methods by correcting only the object's feature points. The proposed algorithm has two steps: first, a local map is created based on an omnidirectional-vision SLAM approach for individual robots. Second, a global map is generated by merging individual maps from multiple robots. The reliability of the proposed mapping algorithm is verified through a comparison of maps based on the proposed algorithm and real maps.

• The Journal of Engineering Geology
• /
• v.32 no.3
• /
• pp.377-388
• /
• 2022

Recently, several studies have utilized machine learning to efficiently and accurately analyze seismic data that are exponentially increasing. In this study, we expand earthquake information such as occurrence time, hypocentral location, and magnitude to produce a dataset for applying to machine learning, reducing the dimension of the expended data into dominant features through principal component analysis. The dimensional extended data comprises statistics of the earthquake information from the Global Centroid Moment Tensor catalog containing 36,699 seismic events. We perform data preprocessing using standard and max-min scaling and extract dominant features with principal components analysis from the scaled dataset. The scaling methods significantly reduced the deviation of feature values caused by different units. Among them, the standard scaling method transforms the median of each feature with a smaller deviation than other scaling methods. The six principal components extracted from the non-scaled dataset explain 99% of the original data. The sixteen principal components from the datasets, which are applied with standardization or max-min scaling, reconstruct 98% of the original datasets. These results indicate that more principal components are needed to preserve original data information with even distributed feature values. We propose a data processing method for efficient and accurate machine learning model to analyze the relationship between seismic data and seismic behavior.