Showing papers by "Hyoung Joong Kim published in 2010"
TL;DR: The approach makes use of self-adaptive error based control parameters to alter the training data sequence, evolve the network architecture, and learn the network parameters and avoids the over-training problem and reduces the training time significantly.
124 citations
TL;DR: The EMD-2 is much better than the EMD, and slightly better than 2-EMD when n is 3, 4 and 5, and the way to generate basis vector can be used for the generalization of the n- EMD and E MD-n where n>1.
Abstract: The original exploiting modification direction (EMD) method proposed by Zhang and Wang is a novel data hiding technique which can achieve large embedding capacity with less distortion. The original EMD method can hide (2n+1)-ary numbers by modifying at most one least-significant bit (LSB) of n pixel values. The proposed methods in this paper, 2-EMD and EMD-2, modify at most two pixels of the LSB values. Efficiency of the proposed methods is shown theoretically and through experiments. The 2-EMD and EMD-2 can hide even larger numbers than the EMD with similar distortion under the same conditions. This paper shows that the EMD-2 is much better than the EMD, and slightly better than 2-EMD when n is 3, 4 and 5. The way to generate basis vector can be used for the generalization of the n-EMD and EMD-n where n>1.
72 citations
TL;DR: This paper presents an improved reversible data hiding scheme using upgraded histogram shifting based on sorting the predicted errors, characterized by the algorithm which is able to find the optimum threshold values and manage the location map effectively.
Abstract: To be reversible as a data hiding method, the original content and hidden message should be completely recovered. One important objective of this approach is to achieve high embedding capacity and low distortion. Using predicted errors is very effective for increasing the embedding capacity. Sorting the predicted errors has a good influence on decreasing distortion. In this paper, we present an improved reversible data hiding scheme using upgraded histogram shifting based on sorting the predicted errors. This new scheme is characterized by the algorithm which is able to find the optimum threshold values and manage the location map effectively. Experimental results compared with other methods are presented to demonstrate the superiority of the proposed method.
61 citations
TL;DR: The experimental results show that the proposed method achieves better performance than Mielikainen's pair-wise LSB matching method in terms of distortion and survival probability against steganalysis.
48 citations
TL;DR: The experimental results show that the section-wise strategy can improve the embedding efficiency and the visual quality further than the EMD method, and reduce the possibility of detection.
46 citations
TL;DR: A new technique for automatic detection and removal of blurry pictures is presented and complexity is kept low by applying a Monte-Carlo like technique for the selection of representative image areas and interest points and by implicitly estimating the gradient of the scale-space curve evolution.
Abstract: Even though state-of-the-art digital cameras are equipped with auto-focusing and motion compensation functions, several other factors including limited contrast, inappropriate exposure time and improper device handling can still lead to unsatisfactory image quality such as blurriness. Indeed, blurry images make up a significant percentage of anyone's picture collections. Consequently, an efficient tool to detect blurry images and label or separate them for automatic deletion in order to preserve storage capacity and the quality of image collections is needed. A new technique for automatic detection and removal of blurry pictures is presented. Initially, a set of interest points and local image areas is extracted. These areas are then evolved in time according to the conventional linear scale space. The gradient of the evolution curve through scale is then used to produce a 'blur graph' representing the probability of a picture being blurred or not. Complexity is kept low by applying a Monte-Carlo like technique for the selection of representative image areas and interest points and by implicitly estimating the gradient of the scale-space curve evolution. An exhaustive evaluation of the proposed technique is conducted to validate its performance in terms of detection accuracy and efficiency.
18 citations
01 Oct 2010
TL;DR: This paper presents JPEG steganography method based on hiding data to the stream of ternary coefficients, which has many solutions for hiding necessary data and has better data hiding performance compared to the existing steganographic methods.
Abstract: In this paper we present JPEG steganography method based on hiding data to the stream of ternary coefficients. In the proposed method each nonzero DCT coefficient is converted to the corresponding ternary coefficient. The block of 3m - 1 ternary coefficients is used for hiding m ternary messages by modifying one or two coefficients. Due to higher information density of the ternary coefficients, the proposed method has many solutions for hiding necessary data. Such a big choice enables to choose coefficients with lowest distortion impact. As a result, the proposed methods have better data hiding performance compared to the existing steganographic methods based on hiding data to stream of binary coefficients like matrix encoding (F5) and modified matrix encoding (MME). The proposed methods were tested with steganalysis method proposed by T. Pevny and J. Fridrich. The experimental results show that the proposed method has less detectability compared to MME (modified matrix encoding).
5 citations
TL;DR: A new shape-based indexing and matching scheme of real or rendered 3-D objects for camera view invariant object retrieval is proposed, which includes: i) mirror shape pairing and ii) camera view pruning according to the symmetrical patterns of object shapes.
Abstract: Camera view invariant 3-D object retrieval is an important issue in many traditional and emerging applications such as security, surveillance, computer-aided design (CAD), virtual reality, and place recognition. One straightforward method for camera view invariant 3-D object retrieval is to consider all the possible camera views of 3-D objects. However, capturing and maintaining such views require an enormous amount of time and labor. In addition, all camera views should be indexed for reasonable retrieval performance, which requires extra storage space and maintenance overhead. In the case of shape-based 3-D object retrieval, such overhead could be relieved by considering the symmetric shape feature of most objects. In this paper, we propose a new shape-based indexing and matching scheme of real or rendered 3-D objects for camera view invariant object retrieval. In particular, in order to remove redundant camera views to be indexed, we propose a camera view skimming scheme, which includes: i) mirror shape pairing and ii) camera view pruning according to the symmetrical patterns of object shapes. Since our camera view skimming scheme considerably reduces the number of camera views to be indexed, it could relieve the storage requirement and improve the matching speed without sacrificing retrieval accuracy. Through various experiments, we show that our proposed scheme can achieve excellent performance.
3 citations
01 Oct 2010
TL;DR: An efficient mechanism which can decrease the size of location map considerably for un-overlapping version of the PS method is presented and achieves more embedding capacity and higher PSNR value due to the reduced size of the location map.
Abstract: In this paper, a method to decrease the size of location map for non-overlapping pattern substitution method is presented. Original pattern substitution (PS) method has been proposed by Ho et al.[1] as a reversible watermarking scheme for binary images. They use a pair of two patterns to embed data. Unfortunately, their location map is huge in size. In our method, we propose an efficient mechanism which can decrease the size of location map considerably for un-overlapping version of the PS method. Experiment results show that our method works well on decreasing the size of location map. Comparison results with the original PS method demonstrate that the proposed method achieves more embedding capacity and higher PSNR value due to the reduced size of the location map.
2 citations
29 Oct 2010
TL;DR: A systematic approach to fight legally against various perceptual errors occurred in analyzing photos and to use as forensic evidence is presented.
Abstract: This paper introduces a new approach to photography forensic field. This approach is on a referenced 3D photography forensic study. Two photos were taken by Alice and Bob which are controversially analyzed based on claims from different sides. With using Google map and two photos, we generate a 3D model and simulate whose claim is more to the truth. Throughout the simulations, one perceptual error committed by Alice has been corrected. At the same time, the claim that Alice has made is not possible to be true. Moreover, this paper covers the fact that optical aberration would be the reason of abnormal effects on the photos which is possible to become a mistake to the normal for the abnormal. As a conclusion, this paper presents a systematic approach to fight legally against various perceptual errors occurred in analyzing photos and to use as forensic evidence.
2 citations
06 Apr 2010
TL;DR: The proposed method keeps as good of a compression rate as arithmetic coding, but takes much less time than that, and experiments show that the proposed theory is true.
Abstract: This paper advances a new lossless data compression method utilizing reversible data hiding to reduce compression times significantly and achieve good compression rates. The core idea of the new approach is to transform a longer binary sequence into a shorter n-ary sequence, transform the n-ary sequence into two binary subsequences, and apply a lossless coding algorithm based on the assumption that a shorter sequence obviously takes less time than a longer sequence. Mathematical conditions to show when the theory proposed in this paper is attainable are provided. The proposed method keeps as good of a compression rate as arithmetic coding, but takes much less time than that. Experiments show that the proposed theory is true.
TL;DR: An (n, n) secret sharing scheme that randomizes one shaze after another by executing XOR operations with random seeds derived from an initial seed that is resistant to collusion attacks.
Abstract: A simple secret sharing scheme for secure visual communications is presented in this paper. Secret sharing schemes allow a group of participants at different locations to share a secret (i.e., an image) among them by splitting it into n pieces (“shazes” or “shadows”. In case of the (k, n) secret shazing scheme only a group of k qualified participants among n (where k≤n) can reconstruct the secret. This paper presents an (n, n) secret sharing scheme. This scheme randomizes one shaze after another by executing XOR operations with random seeds derived from an initial seed. This scheme can also be used as an image encryption scheme. This scheme is resistant to collusion attacks.