• 한국멀티미디어학회논문지
• /
• 제23권12호
• /
• pp.1540-1551
• /
• 2020

Aiming at the problem that the existing human behavior recognition algorithm cannot fully utilize the multi-level spatio-temporal information of the network, a human behavior recognition algorithm based on a dense three-dimensional residual network is proposed. First, the proposed algorithm uses a dense block of three-dimensional residuals as the basic module of the network. The module extracts the hierarchical features of human behavior through densely connected convolutional layers; Secondly, the local feature aggregation adaptive method is used to learn the local dense features of human behavior; Then, the residual connection module is applied to promote the flow of feature information and reduced the difficulty of training; Finally, the multi-layer local feature extraction of the network is realized by cascading multiple three-dimensional residual dense blocks, and use the global feature aggregation adaptive method to learn the features of all network layers to realize human behavior recognition. A large number of experimental results on benchmark datasets KTH show that the recognition rate (top-l accuracy) of the proposed algorithm reaches 93.52%. Compared with the three-dimensional convolutional neural network (C3D) algorithm, it has improved by 3.93 percentage points. The proposed algorithm framework has good robustness and transfer learning ability, and can effectively handle a variety of video behavior recognition tasks.

• 한국군사과학기술학회지
• /
• 제20권1호
• /
• pp.40-47
• /
• 2017

To classify the vehicle targets automatically using thermal images, there are usually two main categories of feature extraction method, local and shape feature extraction methods. Since thermal images have less texture information than color images, the shape feature extraction method is useful when the segmentation results are correct. However, if there are some errors in target segmentation, the shape feature may contain some errors, then the classification accuracy can be decreased. To overcome these problems, in this paper, we propose the segmentation reliability estimation method for target classification. The segmentation reliability can be estimated by using the difference information of average intensities and edge energies between the target and the background area. The estimated segmentation reliability is applied in the decision level fusion method of classification results using both shape and local features. Experiment results using the thermal images of the vehicle targets (main battle tank, armored personnel carrier, military truck, and an estate car) show that the proposed classification method and the segmentation reliability estimation method have a good performance in classification accuracy.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제8권1호
• /
• pp.37-44
• /
• 2019

디지털 IT 기술의 발달로 인하여 프린터와 스캐너의 성능이 향상되고 가격이 저렴해지면서 일반인들도 쉽게 접할 수 있게 되었다. 그러나 이에 따른 부작용으로 공문서 및 사문서 위조 등의 범죄들이 쉽게 이루어질 수 있다. 따라서 해당 문서가 어떤 프린터를 사용하여 출력 되었는가를 특정할 수 있다면 수사 범위를 줄이고 용의자를 판별하는데 도움이 된다. 본 논문에서는 프린터 장치 판별을 위하여 딥러닝 모델을 제안한다. 먼저 최근 인식 등에서 범용적으로 활용되는 지역 특징 기반의 컨볼루셔널 뉴널 네트워크를 이용한 프린터 장치 판별 모델을 제안하고, 전역 특징 기반의 처리 과정을 네트워크 모델에 도입함으로 인하여 수렴 속도 및 정확도를 향상한 기법을 제안한다. 제안한 모델의 성능은 8개의 프린터 장치를 활용하여 기존 프린터 판별을 위한 특징 기반 기술과 비교를 수행하였다. 그 결과 제안하는 지역 특징 기반의 모델과 전역 특징 기반의 모델이 각각 97.23% 및 99.98%의 높은 판별 정확도를 달성하였고, 기존 기술들에 비하여 높은 정확도를 갖는 우수성을 보였다.

• 한국멀티미디어학회논문지
• /
• 제21권12호
• /
• pp.1473-1480
• /
• 2018

The extraction and recognition of human motion characteristics need to combine biometrics to determine and judge human behavior in the movement and distinguish individual identities. The so-called biometric technology, the specific operation is the use of the body's inherent biological characteristics of individual identity authentication, the most noteworthy feature is the invariance and uniqueness. In the past, the behavior recognition technology based on the single characteristic was too restrictive, in this paper, we proposed a mixed feature which combined global silhouette feature and local optical flow feature, and this combined representation was used for human action recognition. And we will use the KTH database to train and test the recognition system. Experiments have been very desirable results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권11호
• /
• pp.4534-4555
• /
• 2015

We propose a robust object cataloging method using multiple locally distinct heterogeneous features for aiding image retrieval. Due to challenges such as variations in object size, orientation, illumination etc. object recognition is extraordinarily challenging problem. In these circumstances, we adapt local interest point detection method which locates prototypical local components in object imageries. In each local component, we exploit heterogeneous features such as gradient-weighted orientation histogram, sum of wavelet responses, histograms using different color spaces etc. and combine these features together to describe each component divergently. A global signature is formed by adapting the concept of bag of feature model which counts frequencies of its local components with respect to words in a dictionary. The proposed method demonstrates its excellence in classifying objects in various complex backgrounds. Our proposed local feature shows classification accuracy of 98% while SURF,SIFT, BRISK and FREAK get 81%, 88%, 84% and 87% respectively.

• 한국콘텐츠학회논문지
• /
• 제13권4호
• /
• pp.73-85
• /
• 2013

이 연구는 온라인 사회자본 형성이라는 개념을 이용하여 지역 언론사의 웹사이트와 소셜미디어 이용을 분석하였다. 언론사의 온라인 사회자본을 웹1.0으로서 홈페이지와 웹2.0으로서 소셜미디어로 나누어 대응 분석을 통해 시각화하였다. 즉 홈페이지에 나타난 웹 피쳐를 분석하고, 소셜미디어의 소셜피쳐와 대표성을 갖는 트위터 이용의 네트워크 구조를 검토하는 것이 목적이다. 온라인 사회자본으로서 웹사이트는 커뮤니케이션, 정보제공, 비즈니스의 측면에서 역할을 하였다. 영남지역 언론사들의 웹페이지 메인에 소셜미디어를 이용해 각각 다른 형태로 웹 피쳐를 통해 네트워크를 확장하려고 하였다. 또한 중앙언론사가 모든 플랫폼을 균형 있게 이용한 반면, 영남언론사는 트위터에 치중되어 있었으며 트위터, 유튜브, 페이스북 순서로 소셜피쳐를 활용하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제5권4호
• /
• pp.840-855
• /
• 2011

In the context of additive noise steganography model, we propose a method to detect least significant bit (LSB) matching steganography in grayscale images. Images are decomposed into detail sub-bands with local linear transform (LLT) masks which are sensitive to embedding. Novel normalized characteristic function features weighted by a bank of band-pass filters are extracted from the detail sub-bands. A suboptimal feature set is searched by using a threshold selection algorithm. Extensive experiments are performed on four diverse uncompressed image databases. In comparison with other well-known feature sets, the proposed feature set performs the best under most circumstances.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제14권3호
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.

• 반도체디스플레이기술학회지
• /
• 제12권2호
• /
• pp.57-61
• /
• 2013

This paper presents adaptive neuro-fuzzy inference based defect detection method for various defect types, such as micro-crack, fingerprint and contamination, in heterogeneously textured surface of polycrystalline solar wafers. Polycrystalline solar wafer consists of various crystals so the surface of solar wafer shows heterogeneously textures. Because of this property the visual inspection of defects is very difficult. In the proposed method, we use local binary feature and fuzzy reasoning for defect detection. Experimental results show that our proposed method achieves a detection rate of 80%~100%, a missing rate of 0%~20% and an over detection (overkill) rate of 9%~21%.

• Journal of Information Processing Systems
• /
• 제13권6호
• /
• pp.1628-1639
• /
• 2017

The combination texture feature extraction approach for texture image retrieval is proposed in this paper. Two kinds of low level texture features were combined in the approach. One of them was extracted from singular value decomposition (SVD) based dual-tree complex wavelet transform (DTCWT) coefficients, and the other one was extracted from multi-scale local binary patterns (LBPs). The fusion features of SVD based multi-directional wavelet features and multi-scale LBP features have short dimensions of feature vector. The comparing experiments are conducted on Brodatz and Vistex datasets. According to the experimental results, the proposed method has a relatively better performance in aspect of retrieval accuracy and time complexity upon the existing methods.