• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1087-1098
• /
• 2023

The rapid advancement of artificial intelligence (AI) technology has led to its proactive utilization across various fields. However, this widespread adoption of AI-based systems has raised concerns about the increasing threat of attacks on these systems. In particular, deep neural networks, commonly used in deep learning, have been found vulnerable to adversarial attacks that intentionally manipulate input data to induce model errors. In this study, we propose a method to protect image classification models from visually imperceptible One-Pixel attacks, where only a single pixel is altered in an image. The proposed defense technique utilizes an autoencoder model to remove potential threat elements from input images before forwarding them to the classification model. Experimental results, using the CIFAR-10 dataset, demonstrate that the autoencoder-based defense approach significantly improves the robustness of pretrained image classification models against One-Pixel attacks, with an average defense rate enhancement of 81.2%, all without the need for modifications to the existing models.

• International journal of advanced smart convergence
• /
• v.9 no.3
• /
• pp.118-126
• /
• 2020

In this paper, we present an effective single image fog removal by using the Joint Bright and Dark Channel (JBDC) and pixel-based transmission estimation to enhance the visibility of outdoor images susceptible to degradation due to weather and environmental conditions. The conventional methods include refinement process of coarse transmission with heavy computational complexity. The proposed transmission estimation reveals excellent edge-preserving performance and does not require the refinement process. We estimate the atmospheric light in pixel-based fashion, which can improve the transmission estimation performance and visual quality of the restored image. Moreover, we propose an adaptive transmission estimation to enhance the visual quality specifically in sky regions. Comprehensive experiments on various fog images show that the proposed method exhibits reduced computational complexity and excellent fog removal performance, compared with the existing methods; thus, it can be applied to various fields including real-time devices.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2004.08a
• /
• pp.373-376
• /
• 2004

The measurement of diffraction of LCOS microdisplay with various pixel size, interpixel gap, pixel height and coatings demonstrates that pixel size is the leading factor for diffraction loss, while the role of varying pixel gap is less significant comparatively. One-dimensional diffraction simulation is found to be in good agreement with the measurement. Noticeable deviation occurs when pixel size is as small as 8um.

• Journal of Advanced Marine Engineering and Technology
• /
• v.28 no.4
• /
• pp.587-592
• /
• 2004

Pixel classification is one of basic issues of image processing. The general characteristics of the pixels belonging to various classes are discussed and the radical principles of pixel classification are given. At the same time, a pixel classification scheme based on image information scales is proposed. The proposed method is overcome that computation amount become greater and contents easily get turned. And image directional scales has excellent anti-noise performance. In the result of experiment. good efficiency is showed compare with other methods.

• Proceedings of the Korean Nuclear Society Conference
• /
• 1998.05b
• /
• pp.705-710
• /
• 1998

We describe a multigrid wavelet-based natural pixel (WNP) method for image reconstruction in emission computed tomography (ECT). The ECT is used to identify the tagged radioactive material's position in the body for detection of abnormal tissue such as tumor or cancer, as in SPECT and PET. With ECT methodology in parallel beam mode, we formulate a matrix-based reconstruction method for radionuclide sources in the human body. The resulting matrix for a practical problem is very large and nearly singular. To overcome this ill-conditioning, wavelet transform is considered in this study. Wavelets have inherent de-noising and multiscale resolution properties. Therefore, the multigrid wavelet-based natural pixel (WNP) method is very efficient to reconstruct image from projection data that is noisy and incomplete. We test this multigrid wavelet natural pixel (WNP) reconstruction method with the MCNP generated projection data for diagnosis of the simulated cancerous tumor.

• Journal of the Semiconductor & Display Technology
• /
• v.8 no.3
• /
• pp.39-44
• /
• 2009

Local Binary Pattern (LBP) is becoming a popular tool for various machine vision applications such as face recognition, classification and background subtraction. In this paper, we propose a new extension of LBP, called the Diagonal LBP (DLBP), to handle the image-based gender classification problem arise in interactive display systems. Instead of comparing neighbor pixels with the center pixel, DLBP generates codes by comparing a neighbor pixel with the diagonal pixel (the neighbor pixel in the opposite side). It can reduce by half the code length of LBP and consequently, can improve the computation complexity. The Support Vector Machine is utilized as the gender classifier, and the texture profile based on DLBP is adopted as the feature vector. Experimental results revealed that our approach based on the diagonal LPB is very efficient and can be utilized in various real-time pattern classification applications.

• ETRI Journal
• /
• v.40 no.5
• /
• pp.677-686
• /
• 2018

A novel data hiding technique based on modified histogram shifting that incorporates multi-bit secret data hiding is proposed. The proposed technique divides the image pixel values into embeddable and nonembeddable pixel values. Embeddable pixel values are those that are within a specified limit interval surrounding the peak value of an image. The limit interval is calculated from the number of secret bits to be embedded into each embeddable pixel value. The embedded secret bits can be perfectly extracted from the stego image at the receiver side without any overhead bits. From the simulation, it is found that the proposed technique produces a better quality stego image compared to other data hiding techniques, for the same embedding rate. Since the proposed technique only embeds the secret bits in a limited number of pixel values, the change in the visual quality of the stego image is negligible when compared to other data hiding techniques.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.43 no.1
• /
• pp.16-25
• /
• 2020

Vegetation segmentation in a field color image is a process of distinguishing vegetation objects of interests like crops and weeds from a background of soil and/or other residues. The performance of the process is crucial in automatic precision agriculture which includes weed control and crop status monitoring. To facilitate the segmentation, color indices have predominantly been used to transform the color image into its gray-scale image. A thresholding technique like the Otsu method is then applied to distinguish vegetation parts from the background. An obvious demerit of the thresholding based segmentation will be that classification of each pixel into vegetation or background is carried out solely by using the color feature of the pixel itself without taking into account color features of its neighboring pixels. This paper presents a new pixel-based segmentation method which employs a multi-layer perceptron neural network to classify the gray-scale image into vegetation and nonvegetation pixels. The input data of the neural network for each pixel are 2-dimensional gray-level values surrounding the pixel. To generate a gray-scale image from a raw RGB color image, a well-known color index called Excess Green minus Excess Red Index was used. Experimental results using 80 field images of 4 vegetation species demonstrate the superiority of the neural network to existing threshold-based segmentation methods in terms of accuracy, precision, recall, and harmonic mean.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.11
• /
• pp.4537-4556
• /
• 2020

This research was based on the study conducted by Khodaei et al. (2012), namely, the least-significant-bit (LSB) substitution combined with the pixel-value differencing (PVD) steganography, and presented an improved irreversible image steganography method. Such a method was developed through integrating the improved LSB substitution with the modulus function-based PVD steganography to increase steganographic capacity of the original technique while maintaining the quality of images. It partitions the cover image into non-overlapped blocks, each of which consists of 3 consecutive pixels. The 2nd pixel represents the base, in which secret data are embedded by using the 3-bit LSB substitution. Each of the other 2 pixels is paired with the base respectively for embedding secret data by using an improved modulus PVD method. The experiment results showed that the method can greatly increase steganographic capacity in comparison with other PVD-based techniques (by a maximum amount of 135%), on the premise that the quality of images is maintained. Last but not least, 2 security analyses, the pixel difference histogram (PDH) and the content-selective residual (CSR) steganalysis were performed. The results indicated that the method is capable of preventing the detection of the 2 common techniques.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2003.07a
• /
• pp.468-471
• /
• 2003

We firstly report liquid crystal shutter based on plastic film and its application to 3D shutter for stereoscopic displays. Plastic liquid crystal shutters have remarkable advantages compared to conventional glass liquid crystal shutters. They are thin, light and non-breakable so that very comfortable 3D shutter eye-wear can be realized using them. The concepts, optical performances and reliability test results of plastic film liquid crystal shutters are presented.