• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2012.10a
• /
• pp.338-340
• /
• 2012

In images, the edge is an important element to analyze characteristics of the image and has been used selectively at several applications. Even now, many researches to detect and take advantage of theses edges are underway and in initially to detect edges, methods using the relation of adjacent pixels are proposed. Characteristic of these methods is that the processing speed of the algorithms is fast, but the specific weighted values are applied to all the pixels regardless of the images equally. In recent years, the research of the edge detection algorithm to adapt according to the image has been actively underway, in order to complement the drawbacks of the existing methods. Therefore, in order to detect the edge excellent characteristics In this paper, we proposed algorithm using adaptive mask.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.17 no.6
• /
• pp.622-632
• /
• 2004

We have studied how liquid crystal's rubbing direction and voltage driving method affect generation of disclination line. At first, generation of disclination line in vertical alignment(VA) cell and VA-twisted nematic(TN) cell has been examined. When liquid crystal's rubbing direction of bottom substrate was 0$^{\circ}$, the degree of generated disclination line was the smallest value. Further, the generation of disclination line above the electrode is less in the frame inversion than in the line inversion. Secondly, we have examined a generation of disclination line in reflective fringe-field switching cell. When the distance between common electrodes is over 3 ${\mu}{\textrm}{m}$ with on-state of one pixel and off-state of neighboring pixels, the reflectance appears only on-state pixel without generating reflectance in adjacent pixels.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.9
• /
• pp.2161-2166
• /
• 2015

This paper analyzes the straight line detection using the principal component analysis (PCA) and proposes its improved algorithm to which two new functions are added. The first function removes invalid pixels through the detected straight line and detects a line again. The second function detects lines from non-overlapped blocks, selects valid line candidates, and detects a valid line from pixels adjacent to each line candidate. The proposed algorithm detects a more accurate straight line with a low computation in comparison with the conventional algorithm in an image with somewhat refined lines.

• Annual Conference of KIPS
• /
• 2019.05a
• /
• pp.515-518
• /
• 2019

Semantic segmentation has remained as a challenging problem in the field of computer vision. Given the immense power of Convolution Neural Network (CNN) models, many complex problems have been solved in computer vision. Semantic segmentation is the challenge of classifying several pixels of an image into one category. With the help of convolution neural networks, we have witnessed prolific results over the time. We propose a convolutional neural network model which uses Fully CNN with deconvolutional pixel layers. The goal is to create a hierarchy of features while the fully convolutional model does the primary learning and later deconvolutional model visually segments the target image. The proposed approach creates a direct link among the several adjacent pixels in the resulting feature maps. It also preserves the spatial features such as corners and edges in images and hence adding more accuracy to the resulting outputs. We test our algorithm on Karlsruhe Institute of Technology and Toyota Technologies Institute (KITTI) street view data set. Our method achieves an mIoU accuracy of 92.04 %.

• Korean Journal of Remote Sensing
• /
• v.22 no.5
• /
• pp.389-396
• /
• 2006

Surface Solar Insolation is important for vegetation productivity, hydrology, crop growth, etc. In this study, Surface Solar Insolation is estimated using Multi-functional Transport Satellite (MTSAT-1R) in clear and cloudy conditions. For the Cloudy sky cases, the surface solar insolation is estimated by taking into account the cloud transmittance and multiple scattering between cloud and surface. This model integrated Kawamura's model and SMAC code computes surface solar insolation with a $5\;km{\times}5\;km$ spatial resolution in hourly basis. The daily value is derived from the available hourly Surface Solar Insolation, independently for every pixel. To validation, this study uses ground truth data recorded from the pyranometer installed by the Korea Meteorological Agency (KMA). The validation of estimated value is performed through a match-up with ground truth. Various match-up with ground truth. Various match-up window sizes are tested with $3{\times}3,\;5{\times}5,\;7{\times}7,\;9{\times}9,\;10{\times}10,\;11{\times}11,\;13{\times}pixels to define the spatial representativity of pyranometer measurement, and to consider drifting clouds from adjacent pixels across the ground station during the averaging interval of 1 hour are taken into account.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2017.04a
• /
• pp.157-157
• /
• 2017

To minimize the damage by wild birds and acquire the benefits such as protection against weeds and maintenance of water content in soil, the mulching black color vinyl after seeding should be carried out. Non-contact and non-destructive methods that can continuously determine the locations are necessary. In this study, a crop position detection method was studied that uses infrared thermal image sensor to determine the cotyledon position under vinyl mulch. The moving system for acquiring image arrays has been developed for continuously detecting crop locations under plastic mulching on the field. A sliding mechanical device was developed to move the sensor, which were arranged in the form of a linear array, perpendicular to the array using a micro-controller integrated with a stepping motor. The experiments were conducted while moving 4.00 cm/s speed of the IR sensor by the rotational speed of the stepping motor based on a digital pulse width modulation signal from the micro-controller. The acquired images were calibrated with the spatial image correlation. The collected data were processed using moving averaging on interpolation to determine the frame where the variance was the smallest in resolution units of 1.02 cm. Non-linear integral interpolation was one of method for analyzing the frequency using the normalization image and then arbitrarily increasing the limited data value of $16{\times}4pixels$ in one frame. It was a method to relatively reduce the size of overlapping pixels by arbitrarily increasing the limited data value. The splitted frames into 0.1 units instead of 1 pixel can propose more than 10 times more accurate and original method than the existing correction method. The non-integral calibration method was conducted by applying the subdivision method to the pixels to find the optimal correction resolution based on the first reversed frequency. In order to find a correct resolution, the expected location of the first crop was indicated on near pixel 4 in the inversion frequency. For the most optimized resolution, the pixel was divided by 0.4 pixel instead of one pixel to find out where the lowest frequency exists.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.22 no.3
• /
• pp.287-293
• /
• 2012

This paper presents an efficient image registration method by measuring the similarity, which is based on multi-dimensional intensity fluctuation. Multi-dimensional intensity which considers 4 directions of the image, is applied to reflect more properties in similarity decision. And an intensity fluctuation is also applied to measure comprehensively the similarity by considering a change in brightness between the adjacent pixels of image. The normalized cross-correlation(NCC) is calculated by considering an intensity fluctuation to each of 4 directions. The 5 correlation coefficients based on the NCC have been used to measure the registration, which are total NCC, the arithmetical mean and a simple product on the correlation coefficient of each direction and on the normalized correlation coefficient by the maximum NCC, respectively. The proposed method has been applied to the problem for registrating the 22 face images of 243*243 pixels and the 9 person images of 500*500 pixels, respectively. The experimental results show that the proposed method has a superior registration performance that appears the image properties well. Especially, the arithmetical mean on the correlation coefficient of each direction is the best registration measure.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.6
• /
• pp.65-71
• /
• 2017

Digital images and videos are subject to various types of noise during storage and transmission. Among these noises, Salt & Pepper noise degrades the compression efficiency of the original data and causing deterioration of performance in edge detection or segmentation used in an image processing method. In order to mitigate this noise, there are many filters such as Median Filter, Weighted Median Filter, Center Weighted Median Filter, Switching Weighted Median Filter and Adaptive Median Filter. However these methods are inferior in performance at high noise density. In this paper we propose a new type of filter for noise mitigation in wireless communication environment where Salt & Pepper noise occurs. The proposed filter detects the location of the damaged pixel by Salt & Pepper noise detection and mitigates the noise by using adjacent pixel values which are not damaged in a certain area. Among the proposed filters, the performance of the filter using the $3{\times}3$ error mask is compared with that of the conventional methods and it is confirmed that when density of noise in the image is 95%, their performances are improved as 13.24 dB compared to MF and 13.09 dB compared to AMF.

• Atmosphere
• /
• v.14 no.2
• /
• pp.11-22
• /
• 2004

A technique on atmospheric correction algorithm to the multi-band reflectance of Landsat TM imagery has been developed using an atmospheric radiation transfer model for eliminating the atmospheric and surface diffusion effects. Despite the fact that the technique of satellite image processing has been continually developed, there is still a difference between the radiance value registered by satellite borne detector and the true value registered at the ground surface. Such difference is caused by atmospheric attenuations of radiance energy transfer process which is mostly associated with the presence of aerosol particles in atmospheric suspension and surface irradiance characteristics. The atmospheric reflectance depend on atmospheric optical depth and aerosol concentration, and closely related to geographical and environmental surface characteristics. Therefore, when the effects of surface diffuse and aerosol reflectance are eliminated from the satellite image, it is actually corrected from atmospheric optical conditions. The objective of this study is to develop an algorithm for making atmospheric correction in satellite image. The study is processed with the correction function which is developed for eliminating the effects of atmospheric path scattering and surface adjacent pixel spectral reflectance within an atmospheric radiation model. The diffused radiance of adjacent pixel in the image obtained from accounting the average reflectance in the $7{\times}7$ neighbourhood pixels and using the land cover classification. The atmospheric correction functions are provided by a radiation transfer model of LOWTRAN 7 based on the actual atmospheric soundings over the Korean atmospheric complexity. The model produce the upward radiances of satellite spectral image for a given surface reflectance and aerosol optical thickness.

• International Journal of Advanced Culture Technology
• /
• v.3 no.2
• /
• pp.161-170
• /
• 2015

Mean separated histogram equalization in order to preserve the original mean brightness has been proposed. To provide the minimum mean brightness error after the histogram modification, the input image's histogram is successively divided by the factor of 2 until the mean brightness error is satisfied the defined threshold. Then each divided group or sub-histogram will be independently equalized based on the proportional input mean. To provide the overall minimum mean brightness error, each group will be controlled by adding some certain pixels from the adjacent grey level of the next group for giving its mean near by the corresponding the divided mean. However, it still exists some little error which will be put into the next adjacent group. By successive dividing the original histogram, we found that the absolute mean brightness error is gradually decreased when the number of group is increased. Therefore, the error threshold is assigned in order to automatically dividing the original histogram for obtaining the desired absolute mean brightness error (AMBE). This process will be applied to the color image by treating each color independently.