• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.9
• /
• pp.4386-4404
• /
• 2016

Most existing salient object detection algorithms commonly employed segmentation techniques to eliminate background noise and reduce computation by treating each segment as a processing unit. However, individual small segments provide little information about global contents. Such schemes have limited capability on modeling global perceptual phenomena. In this paper, a novel salient object detection algorithm is proposed based on region merging. An adaptive-based merging scheme is developed to reassemble regions based on their color dissimilarities. The merging strategy can be described as that a region R is merged with its adjacent region Q if Q has the lowest dissimilarity with Q among all Q's adjacent regions. To guide the merging process, superpixels that located at the boundary of the image are treated as the seeds. However, it is possible for a boundary in the input image to be occupied by the foreground object. To avoid this case, we optimize the boundary influences by locating and eliminating erroneous boundaries before the region merging. We show that even though three simple region saliency measurements are adopted for each region, encouraging performance can be obtained. Experiments on four benchmark datasets including MSRA-B, SOD, SED and iCoSeg show the proposed method results in uniform object enhancement and achieve state-of-the-art performance by comparing with nine existing methods.

• Proceedings of the Korean Society for Bioinformatics Conference
• /
• 2005.09a
• /
• pp.57-62
• /
• 2005

Microarray gene expression profiling technology is one of the most important research topics in clinical diagnosis of disease. Given thousands of genes, only a small number of them show strong correlation with a certain phenotype. To identify such an optimal subset from thousands of genes is intractable, which plays a crucial role when classify multiple-class genes express models from tumor samples. This paper proposes an efficient classifier design method to simultaneously select the most relevant genes using an intelligent genetic algorithm (IGA) and design an accurate classifier using Support Vector Machine (SVM). IGA with an intelligent crossover operation based on orthogonal experimental design can efficiently solve large-scale parameter optimization problems. Therefore, the parameters of SVM as well as the binary parameters for gene selection are all encoded in a chromosome to achieve simultaneous optimization of gene selection and the associated SVM for accurate tumor classification. The effectiveness of the proposed method IGA/SVM is evaluated using four benchmark datasets. It is shown by computer simulation that IGA/SVM performs better than the existing method in terms of classification accuracy.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.2
• /
• pp.125-133
• /
• 2009

RDF is widely used as the ontology language for representing metadata on the semantic web. Since ontology models the real-world, ontology changes overtime. Thus, it is very important to detect and analyze changes in knowledge base system. Earlier studies on detecting changes between RDF models focused on the structural differences. Some techniques which reduce the size of the delta by considering the RDFS entailment rules have been introduced. However, inferencing with RDF models increases data size and upload time. In this paper, we propose a new change detection using RDF reasoning that only computes a small part of the implied triples using backward chaining strategy. We show that our approach efficiently detects changes through experiments with real-life RDF datasets.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.39 no.4
• /
• pp.75-84
• /
• 2014

In this paper, we dealt with feature selection problem of large-scale and high-dimensional biological data such as omics data. For this problem, most of the previous approaches used simple score function to reduce the number of original variables and selected features from the small number of remained variables. In the case of methods that do not rely on filtering techniques, they do not consider the interactions between the variables, or generate approximate solutions to the simplified problem. Unlike them, by combining set covering and clustering techniques, we developed a new method that could deal with total number of variables and consider the combinatorial effects of variables for selecting good features. To demonstrate the efficacy and effectiveness of the method, we downloaded gene expression datasets from TCGA (The Cancer Genome Atlas) and compared our method with other algorithms including WEKA embeded feature selection algorithms. In the experimental results, we showed that our method could select high quality features for constructing more accurate classifiers than other feature selection algorithms.

• The Bulletin of The Korean Astronomical Society
• /
• v.41 no.2
• /
• pp.49.1-49.1
• /
• 2016

Asteroid taxonomy dates back to the mid-1970's and is based mostly on broadband photometric and spectroscopic observations in the visible wavelength. Different taxonomic classes have long been characterized by spectral slope shortward of 0.75 microns and the absorption band in 1 micron, the principal components. In this way, taxonomic classes are grouped and divided into four broad complexes; silicates (S), carbonaceous (C), featureless (X), Vestoids (V), and the end-members that do not fit well within the S, C, X and V complexes. The past decade witnessed an explosion of data due to the advent of large-scale asteroid surveys such as SDSS. The classification scheme has recently been expanded with the analysis of the SDSS 4th Moving Object Catalog (MOC 4) data. However, the boundaries of each complex and subclass are rather ambiguously defined by hand. Furthermore, there are only few studies on asteroid taxonomy using Johnson-Cousins filters, and those were conducted on a small number of objects, with significant uncertainties. In this paper, we present our preliminary results for a new taxonomic classification of asteroids using SMASS, Bus and DeMeo (2014) and the SDSS MOC 4 datasets. This classification scheme is simply represented by a triplet of photometric colors, either in SDSS or in Johnson-Cousins photometric systems.

• Proceedings of the KSRS Conference
• /
• 2008.10a
• /
• pp.173-176
• /
• 2008

Lake studies play an important role in water management, ecology, and other environmental issues. Typically, monitoring lake levels is the first step on the lake studies. However, for the Prairie Pothole Region (PPR) of North America having millions of small lakes and potholes, on-site measurement for lake levels is almost impossible with the conventional gage stations. Therefore, we employed Geographic Information System (GIS) and remote sensing approach with the Shuttle Radar Topography Mission data to extract lake levels. Several image processing techniques were used to extract lake levels for January, 2000 as a one-time snapshot which will be useful in historic lake level reconstruction. This study is associated with other remote sensing datasets such as Landsat imagery and Digital Orthophoto Quadrangle (DOQ). In this research, firstly, image processing techniques like FFT filtering, Lee-sigma, masking with Canny Edge Detector, and contouring were tested for lake level estimation. The semi-automated contouring technique was developed to accomplish the bulk processing for large amount of lakes in this region. Also, effectiveness of each method for bulk processing was evaluated.

• Journal of Korea Multimedia Society
• /
• v.23 no.2
• /
• pp.135-145
• /
• 2020

Recently, deep-learning based methods for low-light image enhancement accomplish great success through supervised learning. However, they still suffer from the lack of sufficient training data due to difficulty of obtaining a large amount of low-/normal-light image pairs in real environments. In this paper, we propose an unsupervised learning approach for single low-light image enhancement using the bright channel prior (BCP), which gives the constraint that the brightest pixel in a small patch is likely to be close to 1. With this prior, pseudo ground-truth is first generated to establish an unsupervised loss function. The proposed enhancement network is then trained using the proposed unsupervised loss function. To the best of our knowledge, this is the first attempt that performs a low-light image enhancement through unsupervised learning. In addition, we introduce a self-attention map for preserving image details and naturalness in the enhanced result. We validate the proposed method on various public datasets, demonstrating that our method achieves competitive performance over state-of-the-arts.

• Journal of Information Processing Systems
• /
• v.10 no.3
• /
• pp.443-458
• /
• 2014

An extreme learning machine (ELM) is a recently proposed learning algorithm for a single-layer feed forward neural network. In this paper we studied the ensemble of ELM by using a bagging algorithm for facial expression recognition (FER). Facial expression analysis is widely used in the behavior interpretation of emotions, for cognitive science, and social interactions. This paper presents a method for FER based on the histogram of orientation gradient (HOG) features using an ELM ensemble. First, the HOG features were extracted from the face image by dividing it into a number of small cells. A bagging algorithm was then used to construct many different bags of training data and each of them was trained by using separate ELMs. To recognize the expression of the input face image, HOG features were fed to each trained ELM and the results were combined by using a majority voting scheme. The ELM ensemble using bagging improves the generalized capability of the network significantly. The two available datasets (JAFFE and CK+) of facial expressions were used to evaluate the performance of the proposed classification system. Even the performance of individual ELM was smaller and the ELM ensemble using a bagging algorithm improved the recognition performance significantly.

• Korean Journal of Remote Sensing
• /
• v.36 no.1
• /
• pp.83-92
• /
• 2020

The purpose of this work was to apply geographical information system (GIS) for geostatistical analyzing by selecting a semi-variogram model to quantify the spatial correlation of the population distribution with residential neighborhoods in the both sides of Mosul city. Two hundred and sixty-eight sample sites in 240 ㎢ are adopted. After determining the population distribution with respect to neighborhoods, data were inserted to ArcGIS10.3 software. Afterward, the datasets was subjected to the semi-variogram model using ordinary kriging interpolation. The results obtained from interpolation method showed that among the various models, Spherical model gives best fit of the data by cross-validation. The kriging prediction map obtained by this study, shows a particular spatial dependence of the population distribution with the neighborhoods. The results obtained from interpolation method also indicates an unbalanced population distribution, as there is no balance between the size of the population neighborhoods and their share of the size of the population, where the results showed that the right side is more densely populated because of the small area of residential homes which occupied by more than one family, as well as the right side is concentrated in economic and social activities.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.10
• /
• pp.4080-4097
• /
• 2020

In recent years, convolutional neural network (CNN) has become the primary method for face detection. But its shortcomings are obvious, such as expensive calculation, heavy model, etc. This makes CNN difficult to use on the mobile devices which have limited computing and storage capabilities. Therefore, the design of lightweight CNN for face detection is becoming more and more important with the popularity of smartphones and mobile Internet. Based on the CPU real-time face detector FaceBoxes, we propose a multi-task lightweight face detector, which has low computing cost and higher detection precision. First, to improve the detection capability, the squeeze and excitation modules are used to extract attention between channels. Then, the textual and semantic information are extracted by shallow networks and deep networks respectively to get rich features. Finally, the landmark detection module is used to improve the detection performance for small faces and provide landmark data for face alignment. Experiments on AFW, FDDB, PASCAL, and WIDER FACE datasets show that our algorithm has achieved significant improvement in the mean average precision. Especially, on the WIDER FACE hard validation set, our algorithm outperforms the mean average precision of FaceBoxes by 7.2%. For VGA-resolution images, the running speed of our algorithm can reach 23FPS on a CPU device.