• IEIE Transactions on Smart Processing and Computing
• /
• v.6 no.3
• /
• pp.168-174
• /
• 2017

Videos recorded on surveillance cameras or by car black boxes at night have distorted images due to illumination variation. Therefore, it is difficult to analyze morphological characteristics of objects, and it is limiting to use such distorted images as evidence in traffic accidents. Image restoration is performed by amplifying the brightness of nighttime images using linearized gamma correction to increase their contrast (which destroys visual information) and by minimizing degradation factors caused by irregular traveling.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.54-57
• /
• 2003

In this paper, we propose an algorithm extracting character regions from scenery images. This algorithm works under a severe constraint: each pixel of a result image must be derived from only information of their neighbor pixels. This constraint is very important for a low cost device like a mobile camera. The proposed algorithm is represented by the local and parallel image processing. It has been tested for 100 scenery images. A result shows that the proposed algorithm can extract character regions at a rate of more than 90％. The result was obtained without learning any template images. the algorithm is very useful.

• Proceedings of the KIEE Conference
• /
• 2000.11d
• /
• pp.761-764
• /
• 2000

This paper presents a Fuzzy Neural Network (FNN) system to decide whether or not the right information of lanes can be extracted from gray-level images of road scene. The decision of noisy level of input images has been required because much noises usually deteriorates the performance of feature detection based on image processing and lead to erroneous results. As input parameters to FNN, eight noisiness indexes are constructed from a cumulative distribution function (CDF) and proved the indexes being classifiers of images as the good and the bad corrupted by sources of noise by correlation analysis between input images and the indexes. Considering real-time processing and discrimination efficiency, the proposed FNN is structured by eight input parameters, three fuzzy variables and single output. We conduct much experiments and show that our system has comparable performance in terms of false-positive rates.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.16 no.5
• /
• pp.178-182
• /
• 2007

This study represents the efficient depth measurement for 3-dimensional microsystems using the disparity histogram from stereo images. Implementation of user-friendly Windows program written in C++ involves the various methods for the stereo-image processing in which the minimization of matching-pixel error upon the unique point for stereo images was carried out as a pre-processing method. Even though MPC among various methods was adopted in the present measurement, the resulting measurements seem to require optimizations of the windows sizes and corrections of post-manipulation for stereo images. The present work using Windows program is promising to measure the 3-dimensional depth of micro-system efficiently in implementing the 3-dimensional structure of micro-systems.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2005.05a
• /
• pp.521-524
• /
• 2005

This paper proposes a fusion algorithm which merges multiple vein pattern images into a single image, larger than those images. As a preprocessing step of template matching, during the verification of biometric data such as fingerprint image, vein pattern image of hand, etc., the fusion technique is used to make reference image larger than the candidate images in order to enhance the matching performance. In this paper, a new algorithm, called BSA (Boundary Stitching Algorithm) is proposed, in which the boundary rectilinear parts extracted from the candidate images are stitched to the reference image in order to enlarge its matching space. By applying BSA to practical vein pattern verification system, its verification rate was increased by about 10%.

• Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
• /
• 2003.04a
• /
• pp.181-185
• /
• 2003

Recently, in accordance with supplying high-resolution satellite images which as IKONOS, KVR-1000, and Quick Bird, the use of satellite images have increased in the study which extraction of features from high-resolution satellite images is becoming a new research focus. In this study, using generally involves such as image segmentation, filtering and sobel operator and thinning in image processing for extraction of feature from satellite image. We apply this method to extraction of feature which need to the revision of map from high-resolution IKONOS satellite image data, we verified the capability of extraction of feature and application using satellite image and proposed a plan for the study in the future.

• Journal of Information Processing Systems
• /
• v.16 no.5
• /
• pp.1183-1195
• /
• 2020

Visual saliency detection is an essential task because it is an important part of various vision-based applications. There are many techniques for saliency detection in color images. However, the number of methods for saliency detection in infrared images is limited. In this paper, we introduce a simple approach for saliency detection in infrared images based on the thresholding technique. The input image is thresholded into several Boolean maps, and an initial saliency map is calculated as a weighted sum of the created Boolean maps. The initial map is further refined by using thresholding, morphology operation, and a Gaussian filter to produce the final, high-quality saliency map. The experiment showed that the proposed method has high performance when applied to real-life data.

• Journal of Information Processing Systems
• /
• v.11 no.4
• /
• pp.528-537
• /
• 2015

Text in images is one of the most important cues for understanding a scene. In this paper, we propose a novel approach based on interest points to localize text in natural scene images. The main ideas of this approach are as follows: first we used interest point detection techniques, which extract the corner points of characters and center points of edge connected components, to select candidate regions. Second, these candidate regions were verified by using tensor voting, which is capable of extracting perceptual structures from noisy data. Finally, area, orientation, and aspect ratio were used to filter out non-text regions. The proposed method was tested on the ICDAR 2003 dataset and images of wine labels. The experiment results show the validity of this approach.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.615-618
• /
• 2019

Visual saliency detection is an important part in various vision-based applications. There are a myriad of techniques for saliency detection in color images. However, the number of methods for saliency detection in infrared images is inadequate. In this paper, we introduce a simple approach for pedestrian identification in infrared images using saliency. The input image is thresholded into several Boolean maps, an initial saliency map is then calculated as a weighted sum of created Boolean maps. The initial map is further refined by using thresholding, morphology operation, and Gaussian filter to produce the final, high-quality saliency map. The experiment showed that the proposed method produced high performance results when applied to real-life data.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.472-474
• /
• 2019

Because of the difference in network structure and loss function, Verification and identification models have their respective advantages and limitations for person reidentification (re-ID). In this work, we propose a multi-task network simultaneously computes the identification loss and verification loss for person reidentification. Given a pair of images as network input, the multi-task network simultaneously outputs the identities of the two images and whether the images belong to the same identity. In experiments, we analyze the major factors affect the accuracy of person reidentification. To address the occlusion problem and improve the generalization ability of reID models, we use the Random Erasing Augmentation (REA) method to preprocess the images. The method can be easily applied to different pre-trained networks, such as ResNet and VGG. The experimental results on the Market1501 datasets show significant and consistent improvements over the state-of-the-art methods.