• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.12C
• /
• pp.1652-1659
• /
• 2004

In this paper, we propose the algorithm that automatically measures the height of the object to move on the base plane by using the geometric information. To extract a moving object from real-time images creates the background image and each pixel is modeled by the three values. The extracted region is represented by cardboard model and calculates the coordinate center in the each part. The top and bottom point of an object are extracted by the calculated coordinate center and an iterative computation. The two points, top and bottom, are used for measuring the height. Given the vanishing line of the ground plane, the vertical vanishing point, and at least one reference height in the scene; then the height of any point from the ground may be computed by specifying the image of the point and the image of the vertical intersection with the ground plane at that point. Through a confidence valuation of the height to be measured, we confirmed similar actual height and result in the simulation experiment.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.1C
• /
• pp.8-16
• /
• 2010

This paper presents a new small target detection method using the cross product of the temporal pixels based on the temporal profile in infrared (IR) image sequences. The temporal characteristics of small targets and the various backgrounds are different. A new algorithm classifies target pixels and the background pixels through the hypothesis testing using the cross product of pixels on the temporal profile and predicts the temporal backgrounds based on the results. The small targets are detected by subtracting the predicted temporal background profile from the original temporal profile. For the performance comparison between the proposed algorithm and the conventional algorithms, the receiver operating characteristics (ROC) curves is computed in experiment. Experimental results show that the proposed algorithm has better discrimination and a lower false alarm rate than the conventional methods.

• The Bulletin of The Korean Astronomical Society
• /
• v.37 no.2
• /
• pp.138.2-138.2
• /
• 2012

Cave on the Moon is considered as the most appropriate place for human to live during the frontier lunar exploration. While the lava flows, the outer crust gets cooled and solidified. Then, the empty space is remained inside after lava flow stops. Such empty space is called the lava caves. Those lava tubes on the Earth are formed mostly by volcanic activity. However, the lava tubes on satellite like Moon and planet like Mars without volcanic activity are mostly formed by the lava flow inside of the crater made by large meteorite impact. Some part of lava tube with collapsed ceiling appears as the entrance of the cave. Such area looks like a deep crater so called a pit crater. Four large pit craters with diameter of > 60 m and depth of > 40 m are found without difficulty from Kaguya and LRO mission image archives. However, those are too deep to use as easily accessible human frontier base. Therefore, now we are going to identify some smaller lunar caves with accessible entrances using LRO camera images of 0.5 m/pixel resolution. Earth's lava caves and their entrances are well photographed by surface and aerial camera in immense volume. Thus, if the image data are sorted and archived well, those images can be used in comparison with the less distinct lunar cave and entrance images due to its smaller size. Then, we can identify the regions on the Moon where there exist caves with accessible entrances. The database will be also useful in modeling geomorphology for lunar and Martian caves for future artificial intelligence investigation of the caves in any size.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.2 no.1
• /
• pp.3-10
• /
• 1998

In this paper, a new superimposition scheme using a computer vision system was proposed with 7 pairs of skull and ante-mortem photographs, which were already identified through other tests and DNA fingerprints at the Korea National Institute of Scientific Investigation. At this computer vision system, an unidentified skull was caught by video-camcoder with the MPEG and a ante-mortem photograph was scanned by scanner. These two images were processed and superimposed using pixel processing. Recognition of the individual identification by anatomical references was performed on the two superimposed images. This image processing techniques for the superimposition of skull and ante-morterm photographs simplify used the previous approach taking skull photographs and developing it to the same size as the ante-mortem Photographs. This system using various image Processing techniques on computer screen, a more precise and time-saving superimposition technique could be able to be applied in the area of computer individual identification.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.8B
• /
• pp.1480-1489
• /
• 1999

In this paper, we propose a new post-processing method which can solve a problem of MSDS(Mean Squared Difference of Slope) method. Using that method the blocking artifacts can significantly be reduced without any restriction, which is a major drawback of block-based DCT compression method. In this approach, the OSLD(Overlapped Sub-Laplacian Distribution) of dequantized block boundary pixel difference values is defined and used to categorize each block of an image into one of four types. Those types are also classified into one of two classes: an edge and a non-edge classes. A slope across the block boundary is used to quantify discontinuity of the image. If an absolute estimated quantization error value of a DCT coefficient is greater than the corresponding quantization step size, it is saturated to the step size in the edge class. The proposed post-processing method can improve not only the PSNR value up to 0.1~O.3 dB but visual quality without any constraints determined by ad-hoc manner.

• The KIPS Transactions:PartB
• /
• v.18B no.3
• /
• pp.139-146
• /
• 2011

The color correction is the important process of influencing on the picture quality of the 3D stereoscopic images. Existing colorcorrecting methods handle the processing intensifying a correspondence among a left and right image using a histogram based on any one side. In case of color correction based on a histogram, it is difficult to correct tone of image, because the color temperature is not converted enough. And in this paper, the color temperature correction and color consistency correction is proposed without using histogram. The proposed color correction method by color temperature gives 3 in CIE-${\Delta}E$ for each pixel on the images captured with same illuminants and the conventional gives similar results. For color consistency, the proposed gives 9 in CIE-${\Delta}E$ on the images captured with different illuminants while the conventional gives 18. The proposed method shows better results than the conventional in color consistency processing.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.9
• /
• pp.2121-2128
• /
• 2015

In this paper presents a method of extracting reliable and regions of interest (ROI) in barcode for the purpose of factory automation. backgrounds are separated based on directional components and the characteristics of detected patterns. post-processing is performed on candidate images with analysis of problems caused by blur, rotation and areas of high similarity. In addition, the resizing factor is used to achieve faster calculations through image resizing. The input images contained multiple product or barcode for application to diverse automation environments; a high extraction success rate is accomplished despite the maximum shooting distance of 80 cm. Simulations involving images with various shooting distances gave an ROI detection rate of 100% and a post-processing success rate of 99.3%.

• Korean Journal of Remote Sensing
• /
• v.25 no.4
• /
• pp.375-381
• /
• 2009

In this paper, synthetic aperture radar (SAR) clutter images are simulated based on the extensive analyses for radar backscatter characteristics of various earth surfaces, and the simulated images are compared with measured SAR images. At first, the surface parameters including soil moisture content and surface roughness parameters and other parameters for vegetation canopies are measured for various surfaces. The backscattering coefficients for the surfaces are computed using theoretical and empirical models for surface scattering and the radiative transfer for vegetation-canopy scattering. Then, the digital elevation map (DEM) and land cover map (LCM) are used for the SAR image generation. The SAR impulse response (correlation function) is also employed to simulated reliable SAR images. Finally, the appropriate speckle and texture parameters for various earth surfaces are used for generating the SAR clutter images.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.2
• /
• pp.49-57
• /
• 2012

Conventional fuzzy c-means (FCM) algorithms have achieved a good clustering performance. However, they do not fully utilize the spatial information in the image and this results in lower clustering performance for images that have low contrast, vague boundaries, and noises. To overcome this issue, we propose an enhanced spatial fuzzy c-means (ESFCM) algorithm that takes into account the influence of neighboring pixels on the center pixel by assigning weights to the neighbors in a $3{\times}3$ square window. To evaluate between the proposed ESFCM and various FCM based segmentation algorithms, we utilized clustering validity functions such as partition coefficient ($V_{pc}$), partition entropy ($V_{pe}$), and Xie-Bdni function ($V_{xb}$). Experimental results show that the proposed ESFCM outperforms other FCM based algorithms in terms of clustering validity functions.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.9
• /
• pp.1049-1059
• /
• 2019

We propose a learning-based image segmentation algorithm. Starting from super-pixels, our method learns the probability of merging two regions based on the ground truth made by humans. The learned information is used in determining whether the two regions should be merged or not in a segmentation stage. Unlike exiting learning-based algorithms, we use both local and object information. The local information represents features computed from super-pixels and the object information represent high level information available only in the learning process. The object information is considered as privileged information, and we can use a framework that utilize the privileged information such as SVM+. In experiments on the Berkeley Segmentation Dataset and Benchmark (BSDS 500) and PASCAL Visual Object Classes Challenge (VOC 2012) data set, out model exhibited the best performance with a relatively small training data set and also showed competitive results with a sufficiently large training data set.