• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.49 no.2
• /
• pp.63-70
• /
• 2012

The collision-free flight planning method based on the extended collision map is proposed for cooperation of multiple unmanned aerial vehicles (UAVs) in a common 3-dimensional workspace. First, a UAV is modeled as a sphere, taking its 3-D motions such as rolling into consideration. We assume that after entering the common workspace, the UAVs move along their straight paths until they depart from the workspace, and that the priorities of the UAVs are determined in advance. According to the assumptions, the collision detection problem between two spheres in $R^3$ can be reduced into the collision detection problem between a circle and a line in $R^2$. For convenience' sake and safety, the collision regions are approximated by collision boxes. Using the collision boxes, the entrance times of the UAVs are scheduled for collision avoidance among the UAVs. By this way, all UAVs can move in the common workspace without collisions with one another. For verifying the effectiveness of the proposed flight planning method, the simulation with 12 UAVs is carried out.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.22 no.4
• /
• pp.1-11
• /
• 2019

With the development of remote sensing sensor technology, it has become possible to acquire satellite images with various spectral information. In particular, since the hyperspectral image is composed of continuous and narrow spectral wavelength, it can be effectively used in various fields such as land cover classification, target detection, and environment monitoring. Change detection techniques using remote sensing data are generally performed through differences of data with same dimensions. Therefore, it has a disadvantage that it is difficult to apply to heterogeneous sensors having different dimensions. In this study, we have developed a change detection method applicable to hyperspectral image and high spat ial resolution satellite image with different dimensions, and confirmed the applicability of the change detection method between heterogeneous images. For the application of the change detection method, the dimension of hyperspectral image was reduced by using correlation analysis and principal component analysis, and the change detection algorithm used CVA. The ROC curve and the AUC were calculated using the reference data for the evaluation of change detection performance. Experimental results show that the change detection performance is higher when using the image generated by adequate dimensionality reduction than the case using the original hyperspectral image.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.6
• /
• pp.1397-1407
• /
• 2015

Individual phenotype is mostly influenced by genetic factors as well as the effects of environmental factors. Therefore, adjustment of environmental factors are needed in order to see more clearly the effects of genetic factors that we are interested in gene screening study related to Hanwoo's economic trait. The purpose of this study is to propose new statistical model that was adjusted environmental factor and identify adjustment effect in a superior gene marker screening study for producing high quality Hanwoo. First, statistical model including both genetic factor and environmental factor establishes and adjusted value of economic trait find by removing effect of environmental factor such as age, breeding farm. Finally, we identify superior gene marker combination and compare accuracy by applying MDR to data of before and after adjustment. Economic trait is used C18:1, SFA, MUFA, MS, CWT, BFT and SNP marker is used 6 markers of LPL that were identified as more excellent SNP marker than the others among 49 markers through fatty acid composition and economic trait performance test.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.11
• /
• pp.41-49
• /
• 2021

All application programs, including malware, call the Application Programming Interface (API) upon execution. Recently, using those characteristics, attempts to detect and classify malware based on API Call information have been actively studied. However, datasets containing API Call information require a large amount of computational cost and processing time. In addition, information that does not significantly affect the classification of malware may affect the classification accuracy of the learning model. Therefore, in this paper, we propose a method of extracting a essential feature set after reducing the dimensionality of API Call information by applying various feature selection methods. We used CICAndMal2020, a recently announced Android malware dataset, for the experiment. After extracting the essential feature set through various feature selection methods, Android malware classification was conducted using CNN (Convolutional Neural Network) and the results were analyzed. The results showed that the selected feature set or weight priority varies according to the feature selection methods. And, in the case of binary classification, malware was classified with 97% accuracy even if the feature set was reduced to 15% of the total size. In the case of multiclass classification, an average accuracy of 83% was achieved while reducing the feature set to 8% of the total size.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.4
• /
• pp.40-50
• /
• 2019

In recent years, it is important to visualize an accurate human body model for the low-end system in the medical imaging field where augmented reality technology and virtual reality technology are used. Decreasing the geometry of a model causes a difference from the original shape and considers the difference as an error. So, the error should be minimized while reducing geometry. In this study, the organ areas of a human body in the tomographic images such as CT or MRI is segmented and 3D geometric model is generated, thereby implementing the reconstruction method of multiple resolution level-of-detail model. In the experiment, a virtual reality platform was constructed to verify the shape of the reconstructed model, targeting the spine area. The 3D human body model and patient information can be verified using the virtual reality platform.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.3
• /
• pp.531-539
• /
• 2019

This paper proposes the training features for malware family analysis and analyzes the multi-classification performance of ensemble models. We construct training data by extracting API and DLL information from malware executables and use Random Forest and XGBoost algorithms which are based on decision tree. API, API-DLL, and DLL-CM features for malware detection and family classification are proposed by analyzing frequently used API and DLL information from malware and converting high-dimensional features to low-dimensional features. The proposed feature selection method provides the advantages of data dimension reduction and fast learning. In performance comparison, the malware detection rate is 93.0% for Random Forest, the accuracy of malware family dataset is 92.0% for XGBoost, and the false positive rate of malware family dataset including benign is about 3.5% for Random Forest and XGBoost.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.4
• /
• pp.77-84
• /
• 2012

Face classification has wide applications in intelligent video surveillance, content retrieval, robot vision, and human-machine interface. Pose and expression changes, and arbitrary illumination are typical problems for face recognition. When the face is captured at a distance, the image quality is often degraded by blurring and noise corruption. This paper investigates the efficacy of multi-classifier decision level fusion for face classification based on the photon-counting linear discriminant analysis with two different cost functions: Euclidean distance and negative normalized correlation. Decision level fusion comprises three stages: cost normalization, cost validation, and fusion rules. First, the costs are normalized into the uniform range and then, candidate costs are selected during validation. Three fusion rules are employed: minimum, average, and majority-voting rules. In the experiments, unfocusing and motion blurs are rendered to simulate the effects of the long distance environments. It will be shown that the decision-level fusion scheme provides better results than the single classifier.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.1
• /
• pp.111-120
• /
• 2017

In this paper, we propose novel feature extraction algorithm for classification of seabed sediment. In previous researches, acoustic reflection coefficient has been used to classify seabed sediments, which is constant in terms of frequency. However, attenuation of seabed sediment is a function of frequency and is highly influenced by sediment types in general. Hence, we developed a feature vector by using attenuation variation with respect to frequency. The attenuation variation is obtained by using reflected signal from the second sediment layer, which is generated by broadband chirp. The proposed feature vector has advantage in number of dimensions to classify the seabed sediment over the classical scalar feature (reflection coefficient). To compare the proposed feature with the classical scalar feature, dimension of proposed feature vector is reduced by using linear discriminant analysis (LDA). Synthesised acoustic amplitudes reflected by seabed sediments are generated by using Biot model and the performance of proposed feature is evaluated by using Fisher scoring and classification accuracy computed by maximum likelihood decision (MLD). As a result, the proposed feature shows higher discrimination performance and more robustness against measurement errors than that of classical feature.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.