• Journal of the Korea Society of Computer and Information
• /
• v.25 no.3
• /
• pp.51-55
• /
• 2020

In this paper, an improved FAM is proposed by adopting similarity learning in the existing FAM (Fuzzy Associative Memory) used in image restoration. Image restoration refers to the recovery of the latent clean image from its noise-corrupted version. In serious application like face recognition, this process should be noise-tolerant, robust, fast, and scalable. The existing FAM is a simple single layered neural network that can be applied to this domain with its robust fuzzy control but has low capacity problem in real world applications. That similarity measure is implied to the connection strength of the FAM structure to minimize the root mean square error between the recovered and the original image. The efficacy of the proposed algorithm is verified with significant low error magnitude from random noise in our experiment.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.8
• /
• pp.1773-1778
• /
• 2012

Nowadays, the high quality of image is required with the demand for digital image processing devices is rapidly increasing. But image always damaged by many kinds of noises and it is necessary to remove noise and the denoising becomes one of the most important fields. In many cases image is corrupted by AWGN(additive white Gaussian noise). In this paper, we proposed an improved denoising algorithm with edge preservation. The proposed algorithm averages values processed by spatial weighted filter and self adaptive weighted filter. Then we add the value which is computed by the equation considering variance of mask and the estimated noise variance. Through the experience, the proposed filter performs well on noise suppression and edge preservation properties and improves the image visual quality.

• Journal of the Korean Home Economics Association
• /
• v.43 no.11 s.213
• /
• pp.97-107
• /
• 2005

Ed- the file appears to be corrupted, and in many sections (these are highlighted) I cannot read it. I presume it has been copied from another format, maybe Hangul 2004. The purpose of this study was to determine a method of make-up image extraction from specific paintings and also to create cyber make-up models according to the images. For this study, Georgia O'Keeffee's floral paintings were analyzed and their colors were compiled to make color palettes. This study attempted to approach floral image make-up which applies specific paintings through the digital mode in the manner of computer graphics. The results of this study were as follows: First, we found romantic images, including feminine, lovely and soft images by Y, GY and RP group colors, in 'Two Calla Lillies on Pink'. Second, we found modem images, including urban, up-to-date and cool images by G, GY and B group colors, in 'Blue and Green Music'. Third, we found sexy images, including brilliant, tempting and daring by R, B and G group colors, in 'Music-Pink and Blue'. To summarize, the images of the paintings were similar to those of the make-up models.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4
• /
• pp.70-76
• /
• 1996

In the case of an image corrupted with mixed noise, conventional MMSE filter can not remove such a mixed noise properly, because the impulse moise cause a certain bias of the minimum mean-square error estimate at regions close to outliers. In this paper, we proposed the new method or removal of mixed noise by combining MMSE filtering structure with local multi-windowing method according to directions and with ranked-order method. As a result, the improvement of the image quality with the proposed was obtained between about 9.7 and 35.2 times in the sense of NMSE(normalized mean square errors) evaluation than that of MMSE filter. Also, we could obtain the enhanced image in the mixed noisy image from visual and quantitative aspect.

• Smart Media Journal
• /
• v.8 no.2
• /
• pp.46-57
• /
• 2019

In recent research, deep learning-based methods have achieved state-of-the-art performance in various computer vision tasks. However, these methods are commonly supervised, and require huge amounts of annotated data to train. Acquisition of data demands an additional costly effort, particularly for the tasks where it becomes challenging to obtain large amounts of data considering the time constraints and the requirement of professional human diligence. In this paper, we present a data level synthetic sampling solution to learn from small and imbalanced data sets using Generative Adversarial Networks (GANs). The reason for using GANs are the challenges posed in various fields to manage with the small datasets and fluctuating amounts of samples per class. As a result, we present an approach that can improve learning with respect to data distributions, reducing the partiality introduced by class imbalance and hence shifting the classification decision boundary towards more accurate results. Our novel method is demonstrated on a small dataset of 2789 tomato plant disease images, highly corrupted with class imbalance in 9 disease categories. Moreover, we evaluate our results in terms of different metrics and compare the quality of these results for distinct classes.

• The KIPS Transactions:PartB
• /
• v.11B no.2
• /
• pp.141-148
• /
• 2004

The ground positions and elevations information called DEMs(Digital Elevation Models) which are extracted from the stereo aerial photographs and/or satellite images using image matching method have the natural errors caused by variant environments. This study suggests the method to evaluate DEMs using correlation values between the reference and the target DEMs. This would be strongly helpful for experts to correct these errors. To evaluate the whole area of DEMs in the horizontal and vertical errors, the target cell is matched for each reference cell using the correlation values of these two cells. When the target cell is matched for each reference cell, horizontal and vertical error was calculated so as to help experts to recognize a certain area of DEMs which should be corrected and edited. If the correlation value is low and tile difference in height is high, the target cell will be candidated as changed or corrupted cell. When the area is clustered with those candidated cells, that area will be regarded as changed or corrupted area to be corrected and edited. Using this method, the evaluation for all DEM cells is practicable, the horizontal errors as well as vertical errors is calculable and the changed or corrupted area can be detected more efficiently.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2005.11a
• /
• pp.167-170
• /
• 2005

Recently, filtering methods for attenuating noise while preserving image details are in progress actively. And SM(standard median) filter showed a great performance for noise removal in impulse noise environment but, it caused edge cancellation error. So, variable methods that modified SM(standard median) filter have been proposed, and CWM(center weighted median) filter is representative. Also, there are several methods to improve the efficiency based on min/max operation in term of preserving detail and filtering speed. In this paper, we managed a pixel corrupted by impulsive noise using min/max value of the surrounding band enclosing a pixel, and compared the efficiency with exiting methods in the simulation.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.4
• /
• pp.72-80
• /
• 1996

In this paper, variable windowing mean filter to remove speckle noise and a measure to detect thin edge in ultrasonic images are proposed. Because ultrasonic images are corrupted by speckle noise showing a granular appearance, good edge detection is difficult. As a result, noise removing filter is needed in preprocessing stage. The speckle noise removing filter is based on mean filter whose window size is changed by the ratio of standard deviation to mean for image signal and noise signal in local area. And the measure expressed the difference of means between tow windows is used for detecting thin edge in filtered image. Results show that variable windowing mean filter removes speckle noise effectively, and proposed measure is useful in detecting thin edge.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2011.05a
• /
• pp.333-335
• /
• 2011

In the transmitting process of image processing system, images always be corrupted by impulse noise, especially random-valued impulse noise. So removing the random-valued impulse noise is very important, but it is also one of the most difficult case in image processing. The most famous method is the standard median filter, but at edge, the filter has a special feature which has a tendency to decrease the preserve. As a result, we proposed a filter that detection random-valued impulse noise firstly, next to use efficient method to remove the noise and preserve the details. And through the simulation, we compared with the algorithms and indicated that proposed method significant improvement over many other existing algorithms.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.246-249
• /
• 2009

Due to the additive white Gaussian noise (AWGN), images are often corrupted. In recent days, Bayesian estimation techniques to recover noisy images in the wavelet domain have been studied. The probability density function (PDF) of an image in wavelet domain can be described using highly-sharp head and long-tailed shapes. If a priori probability density function having the above properties would be applied well adaptively, better results could be obtained. There were some frequently proposed PDFs such as Gaussian, Laplace distributions, and so on. These functions model the wavelet coefficients satisfactorily and have its own of characteristics. In this paper, mixture distributions of Gaussian and Laplace distribution are proposed, which attempt to corporate these distributions' merits. Such mixture model will be used to remove the noise in images by adopting Maximum a Posteriori (MAP) estimation method. With respect to visual quality, numerical performance and computational complexity, the proposed technique gained better results.