• Journal of Information Processing Systems
• /
• v.16 no.5
• /
• pp.1105-1112
• /
• 2020

Gait recognition, as a promising biometric, can be used in video-based surveillance and other security systems. However, due to the complexity of leg movement and the difference of external sampling conditions, gait recognition still faces many problems to be addressed. In this paper, an improved convolutional neural network (CNN) based on Gabor filter is therefore proposed to achieve gait recognition. Firstly, a gait feature extraction layer based on Gabor filter is inserted into the traditional CNNs, which is used to extract gait features from gait silhouette images. Then, in the process of gait classification, using the output of CNN as input, we utilize metric learning techniques to calculate distance between two gaits and achieve gait classification by k-nearest neighbors classifiers. Finally, several experiments are conducted on two open-accessed gait datasets and demonstrate that our method reaches state-of-the-art performances in terms of correct recognition rate on the OULP and CASIA-B datasets.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.11
• /
• pp.2299-2301
• /
• 2009

The adaptive IIR filters for active noise control systems are more effective when acoustic feedback exists, but the adaptive IIR filters could be unstable when the filter algorithm is not yet converged. In this paper, auto-secondary path estimation techniques and a stabilizing process for adaptive Multi-Channel Recursive LMS (MCRLMS) filters are developed to improve the stability of multi-channel active noise control systems. Experiments using a TMS320VC33 digital signal processor in a three dimensional enclosure have performed to show the effectiveness of the proposed algorithm.

• Journal of the Semiconductor & Display Technology
• /
• v.16 no.2
• /
• pp.106-109
• /
• 2017

Recently, ancient documents are actively studied and discussed. However, ancient documents has a few problems on interpretation. The antique documents are too fragile to hand over. So, some studies have been carried out using terahertz to read ancient documents without damaging them. Three techniques are necessary to read letters using terahertz. First, PPEX algorithm, which distinguishes pages. Second, TGSI technique, which distinguishes text from paper on a page. Third, CCSC algorithm, which transforms signals to letters. In this paper, we will describe the preprocessing process to facilitate the recognition of letters before applying the post processing as we mentioned above. Histogram equalization, Histogram stretching and the Sobel filter were applied to the preprocessing.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.5
• /
• pp.609-615
• /
• 2020

In the modern society, the use of digital equipment is increasing along with the 4th industrial revolution, and the importance of image and signal processing is increasing. At the same time, research on noise reduction is being actively conducted. In this paper, we propose a switching filter algorithm for random-valued impulse noise cancellation. The proposed algorithm obtains the threshold value by determining the noise level present in the image, and threshold value is compared with the difference between the input pixel value and the reference value, and is used in the weight switching process of the filter. The final output of the filter is estimated by applying a pixel weight and a modified weight median filter according to the switching, and obtains a final output by comparing the estimated value with the input pixel value. To evaluate the performance of the proposed algorithm, we compared it with the existing methods using simulation and PSNR.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.32 no.10
• /
• pp.844-849
• /
• 2008

We present a novel cell deformability monitoring chip based on the digitally measured cell lysis rate which is dependent on the areal strain of the cell membrane. This method offers simple cell deformability monitoring by automated high-throughput testing system. We suggest the filter design considering the areal strain imposed on the cell membrane passing through the filter array having gradually increased orifice length. In the experiment using erythrocytes, we characterized the cell deformability in terms of average fracture areal strain which was $0.24{\pm}0.014\;and\;0.21{\pm}0.002$ for normal and chemically treated erythrocytes, respectively. We also verified that the areal strain of 0.15 effectively discriminates the deformability difference of normal and chemically treated erythrocytes, which can be applied to the clinical situation. We compared the lysis rates and their difference for the samples from different donors and found that the present chips can be commonly used without any calibration process. The experimental results demonstrate the simple structure and high performance of the present cell deformability monitoring chips, applicable to simple and cost-effective cell aging process monitoring.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.12C
• /
• pp.1004-1009
• /
• 2010

ost digital cameras use single sensor array with color filter array to reduce size and cost. However images taken by single sensor array have only one color component per pixel, to obtain a color image missing two color components need to be reconstructed. This reconstructing process is called as demosaicking. This paper propose a new directional demosaicking method and proposed method achieves better image quality with enhanced weighting function. With comparing objective and subjective performance, we show proposed method achieves better performance than the conventional methods.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.18 no.8
• /
• pp.739-746
• /
• 2005

When we design a binary phase holographic optical low-pass filter (HOLF), the phase difference is generally set to be $\pi$ to optimize the diffraction efficiency. However, the phase difference of real HOLF mostly deviate from $\pi$ by the error in the fabrication process. The deviation causes the (0,0)-th order diffracted beam to increase, which results In raising the diffraction efficiency. To study the effects of the phase error on the performance of HOLF, we calculated the MTF of HOLF for various phase differences. The results show that the phase error of 10 $\%$ makes little change in the filtering characteristics of HOLF. Considering the filtering by lens and CCD, the effects of the phase error becomes much smaller. To confirm it experimentally, we fabricated HOLFs for various phase differences. After installing it in a digital camera, we take picture of test targets and observe the Moire fringes and the resolution. The results agree with our prediction.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2017.05a
• /
• pp.346-349
• /
• 2017

Digital image processing is widely used in a variety of areas, and noise elimination is used as the preprocessing in all the image processing processes. Degradation is occurred in the image data due to multiple reasons. Degradation is to add the noise in the image signal, and salt and pepper noise is the representative one to cause degradation. Therefore, image restoration algorithm was proposed to process with histogram weight filter and median filter by the noise density of local mask to restore the damaged image in the salt and pepper noise environment, in this article. In addition, it was compared with the existing methods using peak signal to noise ratio(PSNR) as the objective determination factor of improvement effect.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.397-400
• /
• 2016

In digital images, the addition due to noise occurs in the process of obtaining, saving, and transmitting. For examples of noise, there are salt and pepper noise, Gaussian noise, and composition noise where various noises are mixed. Existing filters have insufficient noise removal characteristics because it uses single filters in composite noise environment. Therefore the study suggested a switching filter that processes with special weighted value and median filter according to local mask salt and pepper noise density when central pixel is damaged by salt and pepper noise, and processes by applying weighted values differently according to standard deviation of local mask when damaged by Gaussian noise.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2004.05b
• /
• pp.78-81
• /
• 2004

As a society has Progressed rapidly toward a highly advanced digital information age, a multimedia communication service for acquisition, transmission and storage of image data as well as voice has being commercialized externally and internally. However, in the process of digitalization or transmission of data, noise is generated by several causes, and researches for eliminating those noises have been continued until now. The mean filter is useful method to remove AWGN (additive white gaussian noise) from degraded image and has excellent low-frequency properties. However, it brings about degradation of high-frequency properties in image. So in this paper we removed noise with mean filters added directional information and minimized degradation of high-frequency properties.