• Journal of applied mathematics & informatics
• /
• v.36 no.3_4
• /
• pp.317-330
• /
• 2018

We first provide how to apply the global preconditioned conjugate gradient ($G{\ell}-PCG$) method with Kronecker product preconditioners to image deblurring problems with nearly separable point spread functions. We next provide a coarse-grained parallel image deblurring algorithm using the $G{\ell}-PCG$. Lastly, we provide numerical experiments for image deblurring problems to evaluate the effectiveness of the $G{\ell}-PCG$ with Kronecker product preconditioner by comparing its performance with those of the $G{\ell}-CG$, CGLS and preconditioned CGLS (PCGLS) methods.

• Journal of radiological science and technology
• /
• v.45 no.1
• /
• pp.1-9
• /
• 2022

In this study, we proposed an image quality control for an automatic exposure control (AEC) of digital radiographic imaging system and tried to analyze the performance of the AEC by various manufacturer. The subjects of the experiment were analyzed for the AEC image quality evaluation using digital radiation generators from four manufacturer such as PHILIPS, GE Healthcare, SAMSUNG Healthcare, DK Medical Solution. We used as materials for the implementation of the image quality evaluation by coins (500 won, KOMSCO, Korea). This study evaluated the performance evaluation of the AEC as image quality and exposure dose (Milliampere-seconds; mAs). The image quality evaluation was tried visual assessment by two radiologic technologists and contrast to noise (CNR) by ImageJ. The exposure dose investigated mAs on digital radiation generators. The radiographic coin images acquired 360 images based on change in the control factors of the AEC, which were kVp, the consistency of field configuration and dominant zone, sensitivity and density. As a result, there was a significant difference in the AEC performance between manufacturer. The CNR by the AEC for each manufacturer showed a difference of up to about 1.9 times. The exposed tube current by the AEC for each manufacturer showed a difference of up to about 5.8 times. It is expected that our proposed evaluation method using coins could be applied as the AEC performance evaluation method in the future.

• Journal of Korea Multimedia Society
• /
• v.21 no.12
• /
• pp.1387-1395
• /
• 2018

Core algorithm of deep learning Convolutional Neural Network(CNN) shows better performance than other machine learning algorithms. However, if there is not sufficient data, CNN can not achieve satisfactory performance even if the classifier is excellent. In this situation, it has been proven that the use of transfer learning can have a great effect. In this paper, we apply two transition learning methods(freezing, retraining) to three CNN models(ResNet-50, Inception-V3, DenseNet-121) and compare and analyze how the classification performance of CNN changes according to the methods. As a result of statistical significance test using various evaluation indicators, ResNet-50, Inception-V3, and DenseNet-121 differed by 1.18 times, 1.09 times, and 1.17 times, respectively. Based on this, we concluded that the retraining method may be more effective than the freezing method in case of transition learning in image classification problem.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.9
• /
• pp.41-47
• /
• 2022

In this paper, we propose a new and simple self-supervised learning method that predicts the middle image of a face image sequence for automatic expression recognition. Automatic facial expression recognition can achieve high performance through deep learning methods, however, generally requires a expensive large data set. The size of the data set and the performance of the algorithm are tend to be proportional. The proposed method learns latent deep representation of a face through self-supervised learning using an existing dataset without constructing an additional dataset. Then it transfers the learned parameter to new facial expression reorganization model for improving the performance of automatic expression recognition. The proposed method showed high performance improvement for two datasets, CK+ and AFEW 8.0, and showed that the proposed method can achieve a great effect.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.5
• /
• pp.31-40
• /
• 2016

In this paper, we analyse the characteristics of the edge image and propose a new entropy coding optimized to the compression of the edge image. The pixel values of the edge image have the Gaussian distribution around '0', and most of the pixel values are '0'. By using this analysis, the Zero Block technique is utilized in spatial domain. And the Intra Prediction Mode of the edge image is similar to the mode of the surrounding blocks or likely to be the Planar Mode or the Horizontal Mode. In this paper, we make use of the MPM technique that produces the Intra Prediction Mode with high probability modes. By utilizing the above properties, we design a new entropy coding method that is suitable for edge image and perform the compression. In case the existing compression techniques are applied to edge image, compression ratio is low and the algorithm is complicated as more than necessity and the running time is very long, because those techniques are based on the natural images. However, the compression ratio and the running time of the proposed technique is high and very short, respectively, because the proposed algorithm is optimized to the compression of the edge image. Experimental results indicate that the proposed algorithm provides better visual and PSNR performance up to 11 times than the JPEG.

• International Journal of Contents
• /
• v.13 no.3
• /
• pp.1-8
• /
• 2017

Rice plant numbers and density are key factors for yield and quality of rice grains. Precise and properly estimated rice plant numbers and density can assure high yield from rice fields. The main objective of this study was to automatically detect and count rice plants using images of usual field condition from an unmanned aerial vehicle (UAV). We proposed an automatic image processing method based on morphological operation and boundaries of the connected component to count rice plant numbers after transplanting. We converted RGB images to binary images and applied adaptive median filter to remove distortion and noises. Then we applied a morphological operation to the binary image and draw boundaries to the connected component to count rice plants using those images. The result reveals the algorithm can conduct a performance of 89% by the F-measure, corresponding to a Precision of 87% and a Recall of 91%. The best fit image gives a performance of 93% by the F-measure, corresponding to a Precision of 91% and a Recall of 96%. Comparison between the numbers of rice plants detected and counted by the naked eye and the numbers of rice plants found by the proposed method provided viable and acceptable results. The $R^2$ value was approximately 0.893.

• The Journal of Korea Robotics Society
• /
• v.15 no.4
• /
• pp.385-391
• /
• 2020

As deep learning technology has been developed and applied to various fields, it is gradually changing from an existing single image based application to a video based application having a time base in order to recognize human behavior. However, unlike 2D CNN in a single image, 3D CNN in a video has a very high amount of computation and parameter increase due to the addition of a time axis, so improving accuracy in action recognition technology is more difficult than in a single image. To solve this problem, we investigate and analyze various techniques to improve performance in 3D CNN-based image recognition without additional training time and parameter increase. We propose a time base ensemble using the time axis that exists only in the videos and an ensemble in the input frame. We have achieved an accuracy improvement of up to 7.1% compared to the existing performance with a combination of techniques. It also revealed the trade-off relationship between computational and accuracy.

• Journal of Biomedical Engineering Research
• /
• v.14 no.1
• /
• pp.63-72
• /
• 1993

In this paper, the effect of background grey levels on the visual perception of target image displayed on CRT monitor has been investigated. The purpose of this study is to investigate the efficacy of CRT monitor as a display medium of image Information especially in medical imaging field. Tllree sets of experiments have been performed in this study : the first was to measure the luminance response of CRT monitor and to find the best fitting equation, and the second was the psychophysical experiment measuring the threshold grey level differences between the target image and the background required for visual discrimination (or various background grey levels, and the third was to develop a visual model that is predictable of the threshold grey level difference measured in the psychophysical experiment. The result of psycophysical experiment shows that the visual perception performance is significantly degraded in the range of grey levels lower than 50, which is turned out due to she low luminance change of CRT monitor in this range while human eye has been adapted lo relatively bright ambient illumination. And it Is also shown in the simulation study using the developed visual model that the dominant factor degrading the visual performance is the reflected light from the monitor surface by ambient light in general illumination condition.

• International Journal of Control, Automation, and Systems
• /
• v.1 no.4
• /
• pp.495-502
• /
• 2003

This paper focuses on the implementation of an efficient tracking method of a moving object using optimal representative blocks by way of a pan-tilt camera. The key idea is derived from the fact that when the image size of a moving object is shrunk in an image frame according to the distance between the mobile robot camera and the object in motion, the tracking performance of a moving object can be improved by reducing the size of representative blocks according to the object image size. Motion estimations using Edge Detection (ED) and Block-Matching Algorithm (BMA) are regularly employed to track objects by vision sensors. However, these methods often neglect the real-time vision data since these schemes suffer from heavy computational load. In this paper, a representative block able to significantly reduce the amount of data to be computed, is defined and optimized by changing the size of representative blocks according to the size of the object in the image frame in order to improve tracking performance. The proposed algorithm is verified experimentally by using a two degree-of- freedom active camera mounted on a mobile robot.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.49 no.3
• /
• pp.8-14
• /
• 2012

Improving the speed of image processing is in great demand according to spread of high quality visual media or massive image applications such as 3D TV or movies, AR(Augmented reality). SIMD computer attached to a host computer can accelerate various image processing and massive data operations. MAMS is a multi-access memory system which is, along with multiple processing elements(PEs), adequate for establishing a high performance pipelined SIMD machine. MAMS supports simultaneous access to pq data elements within a horizontal, a vertical, or a block subarray with a constant interval in an arbitrary position in an $M{\times}N$ array of data elements, where the number of memory modules(MMs), m, is a prime number greater than pq. MAMS-PP4 is the first realization of the MAMS architecture, which consists of four PEs in a single chip and five MMs. This paper presents implementation of image processing algorithms and performance analysis for MAMS-PP16 which consists of 16 PEs with 17 MMs in an extension or the prior work, MAMS-PP4. The newly designed MAMS-PP16 has a 64 bit instruction format and application specific instruction set. The author develops a simulator of the MAMS-PP16 system, which implemented algorithms can be executed on. Performance analysis has done with this simulator executing implemented algorithms of processing images. The result of performance analysis verifies consistent response of MAMS-PP16 through the pyramid operation in image processing algorithms comparing with a Pentium-based serial processor. Executing the pyramid operation in MAMS-PP16 results in consistent response of processing time while randomly response time in a serial processor.