• Journal of Information Science Theory and Practice
• /
• v.2 no.2
• /
• pp.6-15
• /
• 2014

Initial based disambiguation of author names is a common data pre-processing step in bibliometrics. It is widely accepted that this procedure can introduce errors into network data and any subsequent analytical results. What is not sufficiently understood is the precise impact of this step on the data and findings. We present an empirical answer to this question by comparing the impact of two commonly used initial based disambiguation methods against a reasonable proxy for ground truth data. We use DBLP, a database covering major journals and conferences in computer science and information science, as a source. We find that initial based disambiguation induces strong distortions in network metrics on the graph and node level: Authors become embedded in ties for which there is no empirical support, thus increasing their sphere of influence and diversity of involvement. Consequently, networks generated with initial-based disambiguation are more coherent and interconnected than the actual underlying networks, and individual authors appear to be more productive and more strongly embedded than they actually are.

• ETRI Journal
• /
• v.41 no.6
• /
• pp.797-810
• /
• 2019

Image saliency detection is the basis of perceptual image processing, which is significant to subsequent image processing methods. Most saliency detection methods can detect only a single object with a high-contrast background, but they have no effect on the extraction of a salient object from images with complex low-contrast backgrounds. With the prior knowledge, this paper proposes a method for detecting salient objects by combining the boundary contrast map and the geodesics-like maps. This method can highlight the foreground uniformly and extract the salient objects efficiently in images with low-contrast backgrounds. The classical receiver operating characteristics (ROC) curve, which compares the salient map with the ground truth map, does not reflect the human perception. An ROC curve with distance (distance receiver operating characteristic, DROC) is proposed in this paper, which takes the ROC curve closer to the human subjective perception. Experiments on three benchmark datasets and three low-contrast image datasets, with four evaluation methods including DROC, show that on comparing the eight state-of-the-art approaches, the proposed approach performs well.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.12
• /
• pp.571-576
• /
• 2015

Image dehazing has been extensively studied, but the performance evaluation method for dehazing techniques has not attracted significant interest. This paper surveys many existing performance evaluation methods of image dehazing. In order to analyze the reliability of the evaluation methods, synthetic hazy images are first reconstructed using the ground-truth color and depth image pairs, and the dehazed images are then compared with the original haze-free images. Meanwhile we also evaluate dehazing algorithms not by the dehazed images' quality but by the performance of computer vision algorithms before/after applying image dehazing. All the aforementioned evaluation methods are analyzed and compared, and research direction for improving the existing methods is discussed.

• ETRI Journal
• /
• v.37 no.3
• /
• pp.606-616
• /
• 2015

In this paper, we propose a novel method for road recognition using 3D point clouds based on a Markov random field (MRF) framework in unstructured and complex road environments. The proposed method is focused on finding a solution for an analysis of traversable regions in challenging environments without considering an assumption that has been applied in many past studies; that is, that the surface of a road is ideally flat. The main contributions of this research are as follows: (a) guidelines for the best selection of the gradient value, the average height, the normal vectors, and the intensity value and (b) how to mathematically transform a road recognition problem into a classification problem that is based on MRF modeling in spatial and visual contexts. In our experiments, we used numerous scans acquired by an HDL-64E sensor mounted on an experimental vehicle. The results show that the proposed method is more robust and reliable than a conventional approach based on a quantity evaluation with ground truth data for a variety of challenging environments.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.8
• /
• pp.1918-1924
• /
• 2010

This In this paper, we propose an efficient algorithm for motion estimation of the image sequences with luminance fluctuation. For such sequences, conventional motion estimation methods based on the difference of pixel values usually produce the erroneous motion information. The proposed algorithm defines the luminance fluctuation as a linear model with gain and offset parameter, and extracts motion information using gradient and phase of the corresponding local region within consecutive frames. Therefor the method is robust to the luminance change of the frames. We test our algorithm for the ground truth sequence with artificially added luminance change and motion, and real sequences corrupted by the flicker. The results shows that the proposed algorithm outperforms the conventional methods.

• Proceedings of the KSRS Conference
• /
• 2002.10a
• /
• pp.775-779
• /
• 2002

Currently, high-resolution satellite imagery such as KOMPSAT and IKONOS has been tentatively utilized to various types of urban engineering problems such as transportation planning, site planning, and utility management. This approach aims at software development and followed applications of remotely sensed imagery to transportation geographic analysis. At first, GDPA (Gradient Direction Profile Algorithm) and main modules in it are overviewed, and newly implemented results under MS visual programming environment are presented with main user interface, input imagery processing, and internal processing steps. Using this software, road network are automatically generated. Furthermore, this road network is used to transportation geographic analysis such as gamma index and road pattern estimation. While, this result, being produced to do-facto format of ESRI-shapefile, is used to several types of road layers to urban/transportation planning problems. In this study, road network using KOMPSAT EOC imagery and IKONOS imagery are directly compared to multiple road layers with NGI digital map with geo-coordinates, as ground truth; furthermore, accuracy evaluation is also carried out through method of computation of commission and omission error at some target area. Conclusively, the results processed in this study is thought to be one of useful cases for further researches and local government application regarding transportation geographic analysis using remotely sensed data sets.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.5 no.9
• /
• pp.1613-1631
• /
• 2011

Several metrics have been reported in the literature to assess stereo image quality, mostly based on visual attention or human visual sensitivity based distortion prediction with the help of disparity information, which do not consider the combined aspects of human visual processing. In this paper, visual attention and depth assisted stereo image quality assessment model (VAD-SIQAM) is devised that consists of three main components, i.e., stereo attention predictor (SAP), depth variation (DV), and stereo distortion predictor (SDP). Visual attention is modeled based on entropy and inverse contrast to detect regions or objects of interest/attention. Depth variation is fused into the attention probability to account for the amount of changed depth in distorted stereo images. Finally, the stereo distortion predictor is designed by integrating distortion probability, which is based on low-level human visual system (HVS), responses into actual attention probabilities. The results show that regions of attention are detected among the visually significant distortions in the stereo image pair. Drawbacks of human visual sensitivity based picture quality metrics are alleviated by integrating visual attention and depth information. We also show that positive correlation with ground-truth attention and depth maps are increased by up to 0.949 and 0.936 in terms of the Pearson and the Spearman correlation coefficients, respectively.

• Journal of Korea Multimedia Society
• /
• v.20 no.5
• /
• pp.769-781
• /
• 2017

Deep learning networks like Convolutional Neural Networks (CNNs) show successful performances in many computer vision applications such as image classification, object detection, and so on. For implementation of deep learning networks in embedded system with limited processing power and memory, deep learning network may need to be simplified. However, simplified deep learning network cannot learn every possible scene. One realistic strategy for embedded deep learning network is to construct a simplified deep learning network model optimized for the scene images of the installation place. Then, automatic training will be necessitated for commercialization. In this paper, as an intermediate step toward automatic training under fisheye camera environments, we study more precise human localization in fisheye images, and propose an accurate human localization method, Automatic Ground-Truth Labelling Method (AGTLM). AGTLM first localizes candidate human object bounding boxes by utilizing GoogLeNet-LSTM approach, and after reassurance process by GoogLeNet-based CNN network, finally refines them more correctly and precisely(tightly) by applying saliency object detection technique. The performance improvement of the proposed human localization method, AGTLM with respect to accuracy and tightness is shown through several experiments.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.10a
• /
• pp.397-400
• /
• 2010

The changing light condition of scene cause the luminance fluctuation of the captured image sequences. this artifact is called flicker, and would be easily recognized as visually unstable fluctuation. As the flicker degrades the performance of extracting useful information from image sequences, such as motion information or segmentation, it should be correction and linear flicker model. The algorithm model the flicker effects as a linear system with gain and offset parameter and estimates gain parameter with Gamma correction. The flicker reduction is performed by applying these parameters inversely th the ordinal sequences. To show the performance, we test out algorithm th the ground-truth sequences with the artificially added luminance fluctuation and real sequence with object motion.

• Korean Journal of Remote Sensing
• /
• v.16 no.2
• /
• pp.109-116
• /
• 2000

This study was carried out to assess the use of RADARSAT data which is C-band with HH polarization for the rice growth monitoring in Korea. Nine time-series data were taken by shallow incidence angle (standard beam mode 5 or 6) during rice growing season. And then, backscattering coefficients ($\sigma$$^{\circ}$) were extracted by calibration process for comparing with rice growth parameters such as plant height, leaf area index(LAI), and fresh and dry biomass. Field experimental data concerned with rice growth were collected 8 times for the ground truth at the study area, Tangjin, Chungnam, Korea. At the beginning of rice growth, backscattering coefficients were ranged from -l6~-l3dB when rice fields were not covered with rice canopy and flooded. At the maximum vegetative stage of rice, backscattering coefficients of the rice field were the highest ranging from -4.4dB~-3.1dB. The temporal variation of backscattering coefficient($\sigma$$^{\circ}$) in rice field was significant in this study. Backscattering coefficient ($\sigma$$^{\circ}$) of rice field was a little bit lower again after heading stage than before. This results show RADARSAT data is promising for rice monitoring.