Earticle

본 논문에서는 기존 LDP 코드의 문제점을 보완하고 화소의 색상 정보와 밝기 정보, 에지 방향 정보, 그리고 에지 반 응 크기 정보를 포함할 수 있는 새로운 LDP를 제안한다. 제안된 방법은 얼굴색 정보를 포함하기 위해 기존 LDP 코드 를 줄이는 방법을 제안하고 그 결과를 분석하였다. 새로운 LDP 코드는 기존 LDP 코드와 달리 6비트로 표현함으로써 나머지 2비트에 필요로 하는 정보를 포함할 수 있도록 하였으며, 기존 LDP 코드에 비해서 잡음과 환경 변화에 효과적 으로 적응할 수 있도록 하였다. 실험 결과 제안된 LDP 코드는 기존 방법들에 비해 높은 인식률 향상과 얼굴 표정인식 결과에서도 효과적임을 보여주었다.

In this paper, we proposed a new LDP code to solve a previous LDP code's problem and can include a face-colo r information. To include the face-color information, we developed various methods reducing the existing LDP code and analyzed the results. A new LDP code is represented by 6-bits different from the previous LDP code To adapt to a noise and environmental changes effectively and include 2-bits face-color information. The result shows better recognition rates of face and facial-expression than the existing methods effectively.

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.

객체 인식은 이미지와 비디오에서 다양한 객체를 검출하고 식별하는 기술이다. LBP는 조명 변이에 강인하게 동작하는 기술자로써, 객체 인식 분야에 활발하게 사용되고 있다. LBP는 이웃 픽셀의 범위, 비교 연산 후 이웃들의 결합 순서, 그리고 결합의 시작 위치를 고려해야 한다. 특히, LBP의 시작 위치는 "최상위 비트"가 되기 때문에 객체 인식의 성능에 큰 영향을 미치게 된다. 따라서, 본 논문에서는 N개의 시작 위치를 기반으로 N개의 특징 공간에서 각각 입력 데이터와 가장 유사한 데이터를 탐색한다. 각 특징 공간의 서로 다른 결과를 동일한 기준에서 비교할 수 있는 신뢰 척도를 정의하고, 가장 신뢰도 높은 결과를 선택함으로써 객체 인식을 수행한다. 실험 결과에서는 LBP의 시작 위치에 따라 성능의 차이가 있다는 것을 확인했다. 또한, 제안한 방법은 기존 LBP 인식 성능과 대비하여 최대 12.66%의 높은 성능을 나타냈다.

Object recognition is a technology that detects and identifies various objects in images and videos. LBP is a descriptor that operates robustly to illumination variations and is actively used in object recognition. LBP considers the range of neighboring pixels, the order of combining the neighbors after the comparison operation, and the starting position of combining. In particular, the starting position of the LBP becomes the "most significant bit"; it dramatically affects the performance of object recognition. In this paper, based on the N starting positions, the data most similar to the input data are searched in each of the N feature spaces. Object recognition is performed by the confidence measure that can compare different results of each feature space under the same criterion and select the most reliable result. In the experimental results, it was confirmed that there is a difference in performance depending on the starting position of LBP. The proposed method showed a high performance of up to 12.66% compared to the recognition performance of the existing LBP.

Ship detection is an important application of optical remote sensing image processing. Sea-land segmentation is the key step in ship detection. Traditional sea-land segment methods only based on the gray-level information of an image to choose a gray threshold to segment the image; however, it is very difficult to establish a self-adapting mechanism to select a suitable threshold for different images. Thus, the segmentation result is greatly influenced by the threshold chosen for sea-land segmentation. In this paper, we are integrating the LBP feature information to propose a novel sea-land segmentation algorithm. Moreover, a new ship detection method based on our sea-land segmentation algorithm is proposed for optical remote sensing images. The performance of ship detection is measured in terms of precision and false-alarm-rate. Experimental results show that, as compared to minimum error meth-od, the proposed algorithm can decrease the false-alarm-rate from 23.2% to 9.24%. And compared to Otsu method, the proposed algorithm improve the precision from 82.9% to 90.2%.

In this paper, the researchers propose a two-wheeler detection system based on volumetric local binary pattern (VLBP) using projection vector method for vision based intelligent vehicles. A projection method has robustness for rotation invariant and reducing dimensionality for original image. The features of LBP are fast to compute and simple to implement for object recognition and texture classification area. Moreover, the researchers use uniform pattern to remove the noise. This paper suggests that modified LBP method and projection vector having different weighting values according to the local shape and the area in the image. Also the researchers system maintains the simplicity of evaluation of traditional formulation while being more discriminative. The researchers experimental results show that the detection system of a bicycle and motorcycle based on the proposed VLBP features achieve higher detection accuracy rate than the traditional features

Automated offline handwritten character recognition of Devanagari script is a growing area of research in the field of pattern recognition. A new approach for Devanagari handwritten character / digit recognition has been proposed in this paper. This approach employs Uniform Local Binary Pattern (ULBP) operator as the feature extraction method. This operator has great performance in research areas such as texture classification and object recognition, but it has not been used in Devanagari handwritten character/digit recognition problem. The proposed method extracts both local and global features. The proposed method have two steps, in the first step image is preprocessed to remove noise and to convert it to binary image and then resizing it to a fixed size of 48x48. In the second step, ULBP operator is applied to the image to extract global features then input image is divided into 9 blocks, ULBP operator is applied to each block to extract local features. Finally, global and local features are used to train Support Vector Machine(SVM). The proposed method has been tested on large set of handwritten character and numeral database and empirical results reveals that the proposed method yields very good accuracy (98.77%) . To establish the superiority of the proposed method, it has also been compared with the contemporary algorithms. The comparative analysis shows that the proposed method out performs the existing methods.

In this paper, we present a novel face recognition method called Multi-scale block Center-Symmetric Local Binary Pattern (MCS-LBP). The face recognition process mainly consists of three phase: face representation, feature extraction, and classification. However, the most important phase is extraction, in which unique features of the face image are extracted. The Center-Symmetric Local Binary Pattern (CS-LBP) feature can be viewed as a combination of texture-based features and gradient-based features. However, it has less dimensional area; the bit-wise comparison made between two single pixel values is significantly affected by noise and sensitive to image translation and rotation. To address this problem, we present a modified feature called MCS-LBP. Instead of individual pixels, in the modified CS-LBP, the comparison is performed based on average gray values of sub-regions. Hence, it provides more complete representation than the Local Binary Pattern (LBP) and CS-LBP operator. Experiments demonstrate the proposed method.

In this paper, we propose a new representation, called Local Min-Max Binary Pattern (LMin-MaxBP), and apply it to face recognition with single sample per class. The local appearance based methods have been successfully applied to face recognition and achieved state-of-the-art performance. The Local Binary Pattern (LBP) has been proved to be effective for image representation. The motivation for the LMin-MaxBP is to find texture information to cope with the variation due to facial expression and perspective changes as well as reducing the length of the feature vectors in LBP’s histogram to speed up the matching process. Experiments on Yale, ORL and Indian face datasets shows that the proposed approach improves the performance in the scenario of one training sample per person with significant facial expression and perspective variations with large rotation angle up to 180θ.

Smoke detection plays an important role for the early detection of fire. In this paper, we suggest a newly developed method that generated LBPV(Local Binary Pattern Variance)s as special feature vectors from RGB contrast images can be applied to detect smoke using SVM(Support Vector Machine). The proposed method rearranges mean value of the block from each R, G, B channel and its intensity of the mean value. Additionally, it generates RGB contrast image which indicates each RGB channel’s contrast via smoke’s achromatic color. Uniform LBPV, Rotation-Invariance LBPV, Rotation-Invariance Uniform LBPV are applied to RGB Contrast images so that it could generate feature vector from the form of LBP. It helps to distinguish between smoke and non smoke area through SVM. Experimental results show that true positive detection rate is similar but false positive detection rate has been improved, although the proposed method reduced numbers of feature vector in half comparing with the existing method with LBP and LBPV.

Enormous methods have been proposed for the detection and segmentation of blur and non-blur regions of the images. Due to the limited available information about the blur type, scenario and the level of blurriness, detection and segmentation is a challenging task. Hence, the performance of the blur measure operators is an essential factor and needs improvement to attain perfection. In this paper, we propose an effective blur measure based on the local binary pattern (LBP) with the adaptive threshold for blur detection. The sharpness metric developed based on LBP uses a fixed threshold irrespective of the blur type and level which may not be suitable for images with large variations in imaging conditions and blur type and level. Contradictory, the proposed measure uses an adaptive threshold for each image based on the image and the blur properties to generate an improved sharpness metric. The adaptive threshold is computed based on the model learned through the support vector machine (SVM). The performance of the proposed method is evaluated using a well-known dataset and compared with five state-of-the-art methods. The comparative analysis reveals that the proposed method performs significantly better qualitatively and quantitatively against all the methods.

This paper presents a novel face recognition algorithm based on the deep convolution neural network and key point detection jointed local binary pattern methodology to enhance the accuracy of face recognition. We firstly propose the modified face key feature point location detection method to enhance the traditional localization algorithm to better pre-process the original face images. We put forward the grey information and the color information with combination of a composite model of local information. Then, we optimize the multi-layer network structure deep learning algorithm using the Fisher criterion as reference to adjust the network structure more accurately. Furthermore, we modify the local binary pattern texture description operator and combine it with the neural network to overcome drawbacks that deep neural network could not learn to face image and the local characteristics. Simulation results demonstrate that the proposed algorithm obtains stronger robustness and feasibility compared with the other state-of-the-art algorithms. The proposed algorithm also provides the novel paradigm for the application of deep learning in the field of face recognition which sets the milestone for further research.

최근 임베디드 GPU가 OpenCL 프레임워크를 지원하기 시작하면서, 임베디드 GPU도 GPGPU로 수행을 할 수 있게 되었다. 그러나, 임베디드 GPU는 제한된 수의 프로세싱 코어를 탑재하고 있기 때문에, 최적의 성능을 얻기 위해서는 멀티코어 CPU와 임베디드 GPU를 모두 활용해야 한다. 한편, 얼굴인식은 많은 응용에서 중요해지고 있는데, 보다 정확한 인식을 위해서는 실시간에 더 많은 계산을 해야 한다. 본 논문에서는, 널리 알려진 얼굴인식 알고리즘인 지역적 이진패턴 기반의 얼굴 인식기를 OpenCL을 사용하여 임베디드 GPU에서 가속한 결과를 제시한다. ARM Cortex-A15 듀얼코어 CPU와 ARM Mali쿼드코어 GPU가 하나의 칩으로 구성된 Exynos 5250 시스템에서 가속했고, 태스크 병렬성과 데이터 병렬성을 모두 활용한 병렬화 방안들을 살펴보았다. 단일 쓰레드로 수행하는 CPU 구현에 비해서, 수행시간은 최대 7.5배까지 줄어들었고, 이에 따라 에너지 사용량도 최대 7.0배까지 줄어들었다. 본 논문은 OpenCL을 사용하여 지역적 이진 패턴 기반 얼굴인식을 가속한 최초의 논문이자, Mali GPU에서 OpenCL을 사용한 병렬화에 대한 최초의 논문이다.

Recently, embedded Graphics Processing Units (GPUs) have started to support OpenCL framework which enables general-purpose computing on a GPU (GPGPU). However, the current embedded GPUs usually have a limited number of cores. To achieve better performance in such a system, it is essential to utilize both the multi-core CPU and the GPU at the same time. On the other hand, face recognition is becoming popular in many applications, demanding more computation for higher recognition accuracy in real-time. In this paper, we present an efficient Local Binary Pattern (LBP) based face recognizer implemented using OpenCL on an embedded GPU. We explored different parallelization strategies exploiting both task-parallelism and data-parallelism on a heterogeneous embedded system, Exynos 5250: a system with a dual-core ARM Cortex-A15 CPU and a quad-core ARM Mali GPU. The experimental results show 7.5 times speedup compared to a single threaded CPU execution, and 7.0 times smaller energy consumption. To the best of our knowledge, this is the first work that presents the acceleration of LBP-based face recognition using OpenCL, and more importantly the first work that reports the performance of Mali GPU as a GPGPU using OpenCL.

본 논문에서는 주행 중인 차량에서 전방을 향해 장착된 카메라를 통해 입력된 영상에서 측면에 부분적으로 나타나는 차량을 검출하는 방법을 제안한다. 기존 연구에서는 모션 벡터를 이용하여 주변 배경과 관측되는 차량 사이의 모션 벡터 차이를 이용하여 측면 차량을 검출하고 있다. 그러나 모션 벡터를 이용할 경우 정지된 차량이나 전방에서 다가오는 차량의 경우 검출하기 어려운 점이 있다. 이러한 문제를 해결하기 위해 본 논문에서는 모션 벡터를 사용하지 않고 차량 측면 모습에서 특징 정보를 추출하여 SVM 분류기를 통해 측면 차량을 검출하는 방법을 제안한다. 차량 측면 모습의 특징을 뽑기 위해 영상의 밝기 변화에 강인한 국부 이진 패턴을 사용하였고 ROI영역 내에서 차량이 나타나는 위치에 상관없이 차량의 측면 모습을 찾아내기 위해 국부 이진 패턴의 히스토그램을 이용하였다. 실험결과에서는 제안하는 방법이 정지된 차량을 포함하여 88.5%의 정확도로 측면 차량을 검출하는 것을 확인하였다.

본 논문에서는 조명 또는 장면의 갑작스러운 변화에 효과적으로 배경모델링을 하기 위해 국부이진패턴을 이용한 다중 배경모델링 방법을 제안한다. 제안하는 방법은 각 장면에서 독립적인 배경모델을 이용하여 모델 업데이트를 실시한다. 이후 검출된 전경 영역의 비율이 일정 임계치를 넘게 되면 기존의 모델 중 적합한 모델을 찾거나 새로운 모델을 생성하여 현재 배경모델을 대체한다. 이는 배경모델의 성능을 유지하면서 효율적으로 장면의 변화에 바로 대응할 수 있는 장점이 있다. 실험결과에서는 실내조명이 갑작스럽게 변하는 영상과 Pan Tilt Zoom 카메라를 이용한 다중 영상에서 제안한 방법이 효과적으로 동작함을 확인할 수 있었다.

기존의 색상기반 얼굴인식 방법은 조명변화에 민감하며, 위변조의 가능성이 있기 때문에 다양한 산업분야에 적용되기 어려운 문제가 있었다. 본 논문에서는 이러한 문제를 해결하기 위해 깊이 영상을 이용한 지역 이진 패턴(LBP) 기반의 얼굴인식 방법을 제안한다. 깊이 정보를 이용한 얼굴 검출 방법과 얼굴 인식을 위한 특징 추출 및 매칭 방법을 구현하고, 모의실험 결과를 바탕으로 제안된 방식의 인식 성능을 나타낸다.

Conventional Color-Based Face Recognition Methods are Sensitive to Illumination Changes, and there are the Possibilities of Forgery and Falsification so that it is Difficult to Apply to Various Industrial Fields. In This Paper, we propose a Face Recognition Method Based on LBP(Local Binary Pattern) using the Depth Images to Solve This Problem. Face Detection Method Using Depth Information and Feature Extraction and Matching Methods for Face Recognition are implemented, the Simulation Results show the Recognition Performance of the Proposed Method.

In this paper we suggest a new two-wheelers detection algorithm using uniform local binary pattern weighting value for projection vectors. The first, we calculate feature vectors using projection method which has robustness for rotation invariant and reducing dimensionality for each cell from origin image. The second, we applied new weighting values which are calculated by the modified local binary pattern showing the fast compute and simple to implement. This paper applied the Adaboost algorithm to make a strong classification from weak classification. In this experiment, we can get the result that the detection rate of the proposed method is higher than that of the traditional method.

인간의 행동 분석과 인식은 헬스케어, 작업자 안전 등 다양한 분야에서 활용되고 있다. 행동패턴 분석과 인식에서 주로 IMU센서가 사용되며, 행동인식을 위해 딥 러닝과 SVM법, K-NN 분류법, 주파수 분석법 등 다양한 방법이 연구되고 있다. 이러한 방법은 많은 메모리와 연산시간이 요구되어 소형이고 저가인 IoT 디바이스에 적용하기가 어렵다, 따라서 본 연구에서는 저가의 임베디드 시스템에 적용가능한 행동 인식기법을 제안한다. 제안된 기법은 3축 가속도 센서를 이용하여 평균속도와 영-교차율을 산출하고 이를 이미지화하여 LBP로 인간의 행동패턴을 인식하게 한 것이다. 제안된 기법의 성능을 평가하기 위해 모바일 폰 센서에서 수집한 행동 데이터를 이용하였다. 성능 평가에서 91.8%의 인식률을 확인하였다. 이와같이 제안된 알고리즘을 저가의 임베디드 시스템에 적용한다면 매우 효과적일 것이라 사료된다.

Analysis and recognition of human activity are used in various fields such as healthcare and worker safety. In order to recognize human activity, Various methods such as deep learning, SVM, K-NN classification, and frequency analysis are currently being studied. These methods require a lot of memory and computation time. So, it is difficult to apply them to small and low-cost IoT devices. Therefore, this study proposes a method to recognize the human activity for applicable to low-cost embedded systems. The proposed method calculates the average velocity and the zero-crossing ratio using a three-axis acceleration sensor and then images, and it is recognized the human activity pattern using these images and LBP. In the performance evaluation, it is confirmed that the recognition rate of 91.8% and the computation time are small. Therefore, it is very effective if the proposed algorithm is applied to the low cost embedded system.

최근 얼굴 표정 인식에 있어 영상 기반의 방법의 하나로서 ULBP 블록 히스토그램 피쳐와 SVM을 분류기로 사용한 연구가 수행되었다. Ojala 등에 의해 소개된 LBP는 높은 식별력과 조명의 변화에 대한 내구성과 간단한 연산 때문에 영상 인식 분야에 많이 사용되고 있다. 본 논문에서는 ULBP 블록 히스토그램을 계산함에 있어 분할 영역의 이동, 크기 변화에 더하여 미세한 특징 요소를 표현할 수 있도록 $LBP_{8,2}$과 $LBP_{8,1}$를 결합하였다. $LBP_{8,1}$ 660개, $LBP_{8,2}$ 550개의 분할 창으로부터 1210개의 ULBP 히스토그램 피쳐를 추출하고 이로부터 AdaBoost를 이용하여 50개의 약 분류기를 생성하였다. $LBP_{8,1}$와 $LBP_{8,2}$가 결합된 하이브리드 형태의 ULBP 블록 히스토그램 피쳐와 SVM 분류기를 이용함으로써 표정 인식률을 향상시킬 수 있었으며 다양한 실험을 통하여 이를 확인하였다. 본 논문에서 제안한 하이브리드 Boosted ULBP 히스토그램의 경우에 표정의 인식률이 96.3%로 가장 높은 결과를 보였으며 제안한 방법의 우수성을 확인하였다.

Recently, as one of images based methods in facial expression recognition, the research which used ULBP block histogram feature and SVM classifier was performed. Due to the properties of LBP introduced by Ojala, such as highly distinction capability, durability to the illumination changes and simple operation, LBP is widely used in the field of image recognition. In this paper, we combined $LBP_{8,2}$ and $LBP_{8,1}$ to describe micro features in addition to shift, size change in calculating ULBP block histogram. From sub-windows of 660 of $LBP_{8,1}$ and 550 of $LBP_{8,2}$, ULBP histogram feature of 1210 were extracted and weak classifiers of 50 were generated using AdaBoost. By using the combined $LBP_{8,1}$ and $LBP_{8,2}$ hybrid type of ULBP histogram feature and SVM classifier, facial expression recognition rate could be improved and it was confirmed through various experiments. Facial expression recognition rate of 96.3% by hybrid boosted ULBP block histogram showed the superiority of the proposed method.

본 논문에서는 고차원 국부이진패턴과 결합베이시안 알고리즘을 이용한 얼굴인증 임베디드 시스템을 제안한다. 또한, 제안된 알고리즘에 대한 임베디드 시스템을 라즈베리파이 3을 이용하여 구현한 결과를 제시한다. 제안된 얼굴인증 알고리즘에 대한 평가는 500명의 얼굴 데이터가 저장된 데이터베이스를 이용하여 수행하였다. 여기서 각각의 얼굴 데이터는 학습용과 테스트용 이미지로 구성하였다. 성능평가를 위한 척도로는 주성분분석법의 차원에 따른 스코어 분포와 얼굴인증 시간을 이용하였다. 그 결과, 최적화된 임베디드 환경에서 우수한 얼굴인증 성능을 가지는 임베디드 시스템을 상대적으로 저렴한 비용으로 구현할 수 있음을 확인하였다.

In this paper, an embedded system for face authentication, which exploits high-dimensional local binary pattern (LBP) descriptor and joint Bayesian algorithm, is proposed. We also present a feasible embedded system for the proposed algorithm implemented with a Raspberry Pi 3 model B. Computer simulation for performance evaluation of the presented face authentication algorithm is carried out using a face database of 500 persons. The face data of a person consist of 2 images, one for training and the other for test. As performance measures, we exploit score distribution and face authentication time with respect to the dimensions of principal component analysis (PCA). As a result, it is confirmed that an embedded system having a good face authentication performance can be implemented with a relatively low cost under an optimized embedded environment.

PCB 제작 분야에서 TCP와 COF에서는 전기적인 특성검사만으로 자동화를 이루어지고 있으며, 실제 단락 및 돌기(근사단락) 형태의 데이터 불량 등에 대해서는 노동력을 동원해 불량을 검출하고 있는 실정이다. 본 논문에서는 영상처리에 의해 국부지역패턴 분석법에 기초한 검출기법을 제안한다. 제안한 방법은 히스토그램보정, 공간위치보정 및 최대왜곡좌표를 구하는 전처리 과정을 포함하여, 지역기반의 패턴분석법이 적용된다. 모의실험을 통하여 제안한 방식은 기존의 영역기반의 검출기법에 비해 성능이 개방 및 근사개방 결함 검출에서 크게 성능을 개선할 수 있음을 보인다.

Conventionally, PCB fabrication processes detects simply electrical characteristics of TCP and COF by automatic manufacturing system and additionally, by introducing human visual detection, those are very ineffective in view of low cost implementation. So, this paper presents an efficient detection algorithm for short and protrusion-type defects based on reference images by using local binary pattern analysis. The proposed methods include several preprocessing techniques such as histogram equalizing, the compensation of spatial position and maximum distortion coordination Through several experiments, it is shown that the proposed method can improve the defect detection performance compared to the conventional schemes.