• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.3
• /
• pp.565-577
• /
• 2019

In recent years, biometric authentication has been used for various applications. Since biometric features are unchangeable and cannot be revoked unlike other personal information, there is increasing concern about leakage of biometric information. Recently, Jin et al. proposed a new cancelable biometric scheme, called "Index-of-Max" (IoM) to protect fingerprint template. The authors presented two realizations, namely, Gaussian random projection-based and uniformly random permutation-based hashing schemes. They also showed that their schemes can provide high accuracy, guarantee the security against recently presented privacy attacks, and satisfy some criteria of cancelable biometrics. However, the authors did not provide experimental results for other biometric features (e.g. finger-vein, iris). In this paper, we present the results of applying Jin et al.'s scheme to iris data. To do this, we propose a new method for processing iris data into a suitable form applicable to the Jin et al.'s scheme. Our experimental results show that it can guarantee favorable accuracy performance compared to the previous schemes. We also show that our scheme satisfies cancelable biometrics criteria and robustness to security and privacy attacks demonstrated in the Jin et al.'s work.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.25-31
• /
• 2021

As the machine learning becomes more common, development of application using machine learning is actively increasing. In addition, research on machine learning platform to support development of application is also increasing. However, despite the increasing of research on machine learning platform, research on suitable load balancing for machine learning platform is insufficient. Therefore, in this paper, we propose a load balancing scheme that can be applied to machine learning distributed environment. The proposed scheme composes distributed servers in a level hash table structure and assigns machine learning task to the server in consideration of the performance of each server. We implemented distributed servers and experimented, and compared the performance with the existing hashing scheme. Compared with the existing hashing scheme, the proposed scheme showed an average 26% speed improvement, and more than 38% reduced the number of waiting tasks to assign to the server.

• Journal of Korea Multimedia Society
• /
• v.19 no.2
• /
• pp.138-145
• /
• 2016

LLAH(Locally Likely Arrangement Hashing) is a method which describes image features by exploiting the geometric relationship between their neighbors. Inherently, it is more robust to large view change and poor scene texture than conventional texture-based feature description methods. However, LLAH strongly requires that image features should be detected with high repeatability. The problem is that such requirement is difficult to satisfy in real applications. To alleviate the problem, this paper proposes a method that improves the matching rate of LLAH by exploiting together the brightness of features. Then, it is verified that the matching rate is increased by about 5% in experiments with synthetic images in the presence of Gaussian noise.

• Proceedings of the IEEK Conference
• /
• 2002.07a
• /
• pp.62-65
• /
• 2002

The locking is the most widely-used concurrency control mechanism for guaranteeing logical consistency of a database where a number of transactions perform concurrently. In this Paper, we propose a new method for lock management appropriate in main-memory databases. Our method chooses the partition, a fixed-sized container for records. as a unit of locking. and directly keeps lock information within the Partition itself. These make our method enjoy the following advantages: (1) it has freedom in controlling of the trade-off between the system concurrency and the lock processing overhead by considering the characteristics of given target applications. (2) it enhances the overall system performance by eliminating the hashing overhead, a serious problem occurred in the traditional method.

• Journal of Korea Multimedia Society
• /
• v.18 no.8
• /
• pp.901-908
• /
• 2015

Methods of detecting, describing, matching image features, like corners and blobs, have been actively studied as a fundamental step for image processing and computer vision applications. As one of feature description/matching methods, LLAH(Locally Likely Arrangement Hashing) describes image features based on the geometric relationship between their neighbors, and thus is suitable for scenes with poor texture. This paper presents a modified LLAH algorithm, which includes the image features themselves for robustly describing the geometric relationship unlike the original LLAH, and employes a voting-based feature matching scheme that makes feature description much simpler. Then, this paper quantitatively analyzes its performance with synthetic images in the presence of Gaussian noise.

• Proceedings of the IEEK Conference
• /
• 1998.10a
• /
• pp.799-802
• /
• 1998

As we know, gigabit ethernet is a new technology to be substituted for current fast ethernet used widely in local area network. The switch used in gigabit ethernet should deal with frames in giga-bps. To do such a fast switching, we need that serveral processes meet the budgets, such as MAC address table lookup, several giga speed path setup, fast scheduling, and etc. Especially MAC address table lookup has to be processed in the same speed with speed of incoming packets, thus the bottleneck in the process can cause packet loss by the overflow in the input buffer. We devise new practical hardware hashing method to perform fast table lookup by minimizing the number of external memory access and accelerating with hardware.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.246-249
• /
• 2016

본 논문에서는 화면 내 블록 카피 (IntraBC: Intra Block Copy) 예측 기술의 압축 성능 분석과 향상된 해쉬 기법을 통한 HEVC (High Efficiency Video Coding) 스크린 콘텐츠 코딩 성능 기법을 제안한다. 현재 SCM (Screen Content Coding Test Model) 에 채택 된 화면 내 블록 카피 기술에서는 $16{\times}16$ 블록에는 1차원 탐색을 수행하고 $8{\times}8$블록에서는 해쉬기반 전역 탐색을 수행하여 해쉬가 일치하는 블록들과 RD-Cost를 수행한다. 현재의 해쉬기반 전역탐색에는 기울기 (Gradient) 위주의 해쉬 구성으로 인해 해쉬가 고르게 분포하지 않아, RD-Cost 수행횟수가 과도하게 많아지는 문제가 있다. 제안하는 방법은 전역적 화면 내 블록 카피의 해쉬 구성 방법을 개선함으로써, 기존 SCM-6.1 대비 0.46%의 BDBR 향상을 확인하였다.

• Journal of Korea Multimedia Society
• /
• v.23 no.1
• /
• pp.8-14
• /
• 2020

For a reliable fingerprinting system, improving fingerprint matching accuracy is crucial. In this paper, we try to improve a binary image fingerprint matching performance by utilizing auxiliary information, power mask, which is obtained while constructing fingerprint DB. The power mask is an expected robustness of each fingerprint bit. A caveat of the power mask is the increased storage cost of the fingerprint DB. This paper mitigates the problem by reducing the size of the power mask utilizing spatial correlation of an image. Experiments on a publicly-available image dataset confirmed that the power mask is effective in improving fingerprint matching accuracy.

• Journal of KIISE:Databases
• /
• v.27 no.2
• /
• pp.216-226
• /
• 2000

This paper suggests a new access path, hashed B-tree which is an efficient access method for integrity checking. Hashed B-tree is based on the observation that most query patterns in enforcing integrity constraints are point queries. Hashed B-tree compresses the key by hashing procedure, which reduces the height of tree and results in fast node search. This method has the advantages such as it can be implemented easily and use the B-tree concurrency control and recovery algorithm with minor modifications.

• The Journal of Information Technology and Database
• /
• v.8 no.2
• /
• pp.103-114
• /
• 2001

In this paper, we propose an efficient index mechanism for real-time main memory database systems. Existing main memory index structures based on the tree can effectively support range searches. However, it doesn't guarantee the real-time characteristic because difference between the access time of a node and an average access time can be high. The index structures based on the hash have always a regular random access time on the simple searches and that speed is very fast. However they do not support range searches. To solve such problems, we propose a new index mechanism called Hyper Tree-Hash (Hyper-TH) that combines ECBH (Extendible Chained Bucket Hashing) and T＊-tree. ECBH can be dynamically extended and has a very fast access time. T＊-tree effectively supports the range searches. We show through our experiments that the proposed mechanism outperforms existing other index structures.