• Journal of the Optical Society of Korea
• /
• 제15권1호
• /
• pp.15-21
• /
• 2011

This paper proposes a new approach to palmprint verification based on the gradient, in which a palm image is considered to be a three-dimensional terrain. Principal lines and wrinkles make deep and shallow valleys on a palm landscape. Then the steepest slope direction in each local area is first computed using the Kirsch operator, after which an orientation map is created that represents the dominant slope direction of each pixel. In this study, three orientation maps were made with different scales to represent local and global gradient information. Next, feature matching based on pixel-unit comparison was performed. The experimental results showed that the proposed method is superior to several state-of-the-art methods. In addition, the verification could be greatly improved by fusing orientation maps with different scales.

• 한국IT서비스학회지
• /
• 제15권4호
• /
• pp.85-95
• /
• 2016

Recently, deep learning and machine learning have attracted considerable attention and many supporting frameworks appeared. In artificial intelligence field, a large body of research is underway to apply the relevant knowledge for complex problem-solving, necessitating the application of various learning algorithms and training methods to artificial intelligence systems. In addition, there is a dearth of performance evaluation of decision making agents. The decision making agent that can find optimal solutions by using reinforcement learning methods designed through this research can collect raw pixel data observed from dynamic environments and make decisions by itself based on the data. The decision making agent uses convolutional neural networks to classify situations it confronts, and the data observed from the environment undergoes preprocessing before being used. This research represents how the convolutional neural networks and the decision making agent are configured, analyzes learning performance through a value-based algorithm and a policy-based algorithm : a Deep Q-Networks and a Policy Gradient, sets forth their differences and demonstrates how the convolutional neural networks affect entire learning performance when using pixel data. This research is expected to contribute to the improvement of artificial intelligence systems which can efficiently find optimal solutions by using features extracted from raw pixel data.

• Journal of the Optical Society of Korea
• /
• 제16권4호
• /
• pp.372-380
• /
• 2012

The optical method Digital Speckle Correlation Measurement (DSCM) has been extensively applied due its capability to measure the entire displacement field over a body surface. A formula of displacement measurement errors by the gradient-based DSCM method was derived. The errors were found to explicitly relate to the image grayscale errors consisting of sub-pixel interpolation algorithm errors, image noise, and subset deformation mismatch at each point of the subset. A power-law dependence of the standard deviation of displacement measurement errors on the subset size was established when the subset deformation was rigid body translation and random image noise was dominant and it was confirmed by both the numerical and experimental results. In a gradient-based algorithm the basic assumption is rigid body translation of the interrogated subsets, however, this is in contradiction to the real circumstances where strains exist. Numerical and experimental results also indicated that, subset shape function mismatch was dominant when the order of the assumed subset shape function was lower than that of the actual subset deformation field and the power-law dependence clearly broke down. The power-law relationship further leads to a simple criterion for choosing a suitable subset size, image quality, sub-pixel algorithm, and subset shape function for DSCM.

• 디지털콘텐츠학회 논문지
• /
• 제4권2호
• /
• pp.211-221
• /
• 2003

본 논문은 중심 화소값과 인접 가해 성분값의 평균으로 축소 성분값을 산출함으로써 ZOD(Zero Order Decimation)와 FOD(First Order Decimation)의 장점을 적응적으로 반영한 디지털 문서 영상 축소 알고리즘의 성능을 분석함에 그 목적이 있다. 제안된 방법은, 슬라이딩 윈도우의 중앙에 위치되는 중심 화소를 축소 성분값의 주성분으로 선택하고, 1차 미분 연산자를 이용하여 중심 화소의 우측 및 하측 인접 화소의 기울기의 크기를 각각 계산한다. 이후, 두 기울기의 크기를 합산한 결과로 각 기울기의 크기를 나누어 우측 및 하측 인접 화소 각각의 국부 가해 가중치를 구한다. 다음으로, 각각의 국부 가해 가중치를 우측 및 하측 인접 화소값에 곱한 후에 그 결과를 합산함으로써 인접 가해 성분값을 산출한다. 이렇게 구한 인접 가해 성분값과 중심 화소값을 평균하여 축소 성분값을 구하는 과정을 입력 영상의 모든 화소들에 반복적으로 수행함으로써 축소 영상을 얻을 수 있다. 본 논문에서는 주관적인 성능과 하드웨어 복잡도 측면에서 제안된 방법과 기존의 각 방식에 대한 성능을 분석했고. 이러한 분석 결과를 토대로 개선된 디지털 문서 영상 축소 알고리즘을 개발하기 위한 바람직한 접근법에 대해 고찰했다.

• International journal of advanced smart convergence
• /
• 제7권4호
• /
• pp.108-113
• /
• 2018

In this study, we introduce a new image gradient computation based on understanding of image generation. Most images consist of groups of pixels with similar color information because the images are generally obtained by taking a picture of the real world. The general gradient operator for an image compares only the neighboring pixels and cannot obtain information about a wide area, and there is a risk of falling into a local minimum problem. Therefore, it is necessary to attempt to introduce the gradient operator of the interval concept. We present a bow-tie gradient by color values of pixels on bow-tie region of a given pixel. To confirm the superiority of our study, we applied our bow-tie gradient to image segmentation algorithms for various images.

• 전자공학회논문지
• /
• 제51권11호
• /
• pp.66-72
• /
• 2014

Histogram of oriented gradient (HOG) 특징은 영상 기반 보행자 인식에서 널리 사용되고 있다. HOG 특징을 이용한 보행자 인식의 인식률을 높이는데 가장 중요한 역할을 하는 것은 보간 기술이다. HOG 특징 연산에 보간 기술을 적용하기 위해서는 각 픽셀의 기울기 방향에 가장 근접한 두 개의 기울기 방향 bin과 가중치를 계산해야 한다. 따라서 본 논문에서는 HOG 특징 연산에 적용하기 위한 효율적인 기울기 방향 bin 및 가중치 연산 회로를 제안한다. 제안하는 회로는 탄젠트 함수와 나눗셈 연산을 피하기 위해 미리 계산된 값을 테이블로 지정하여 사용하였으며, 탄젠트 함수와 가중치 값의 특성을 이용함으로써 회로 내 테이블의 크기를 최소화하였다. 또한 처리 속도 향상을 위해 파이프라인 구조를 적용하였으며, 효율적인 coarse 및 fine 탐색 방법을 적용하여 각 픽셀에 대한 기울기 방향 bin과 가중치를 두 클락 사이클 내에 계산한다. 본 논문에서 제안하는 회로는 $1^{\circ}$ 단위로 기울기 방향을 계산하여 기울기 방향 bin과 가중치를 모두 결정하기 때문에 HOG 특징을 위한 보간 기술에 적용되어 높은 인식률을 제공하기 위해 사용될 수 있다.

• Journal of Information Processing Systems
• /
• 제9권4호
• /
• pp.575-591
• /
• 2013

Most of the edge detection methods available in literature are gradient based, which further apply thresholding, to find the final edge map in an image. In this paper, we propose a novel method that is based on fuzzy logic for edge detection in gray images without using the gradient and thresholding. Fuzzy logic is a mathematical logic that attempts to solve problems by assigning values to an imprecise spectrum of data in order to arrive at the most accurate conclusion possible. Here, the fuzzy logic is used to conclude whether a pixel is an edge pixel or not. The proposed technique begins by fuzzifying the gray values of a pixel into two fuzzy variables, namely the black and the white. Fuzzy rules are defined to find the edge pixels in the fuzzified image. The resultant edge map may contain some extraneous edges, which are further removed from the edge map by separately examining the intermediate intensity range pixels. Finally, the edge map is improved by finding some left out edge pixels by defining a new membership function for the pixels that have their entire 8-neighbourhood pixels classified as white. We have compared our proposed method with some of the existing standard edge detector operators that are available in the literature on image processing. The quantitative analysis of the proposed method is given in terms of entropy value.

• 한국컴퓨터정보학회논문지
• /
• 제17권2호
• /
• pp.59-67
• /
• 2012

본 논문에서는 실제 영상에 일반적으로 존재하는 지역성과 실제 영상에 존재하는 단순 볼록 곡면 특성과 단순 오목 곡면의 특성을 충실히 반영하도록 확대 영상의 보간 픽셀 값들을 추정하기 위하여 인접 픽셀 값과의 기울기 정보를 이용하는 효율적인 보간 기법을 제안하였다. 제안된 기법을 적용하여 확대한 영상의 화질 향상을 측정하기 위하여 PSNR(Peak Signal to Noise Ratio)을 사용하였다. 제안된 기법을 적용하여 확대한 다양한 영상들의 PSNR 값들이 기존의 보간 기법들을 적용하여 확대한 영상들의 PSNR 값보다 큰 것을 확인하였다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제4권2호
• /
• pp.155-160
• /
• 2004

In this paper, we describe the results of a study on the edge enhancement of X-ray images by using their fuzzy system representation. A set of gray scale X-ray images was generated using the EGS4 computer code. An aluminum plate or a lead plate with three parallel strips taken out has been used as the object with the thickness and the width of the plate, and the gap between the two strips varied. We started with a comparative study on a set of the fuzzy sets for their applicability as the input fuzzy sets for the fuzzy system representation of the gray scale images. Then we describe how the fuzzy system is used to sharpen the edges. Our algorithm is based on adding the magnitude of the gradient not to the pixel value of concern but rather to the nearest neighboring pixel in the direction of the gradient. We show that this algorithm is better in maintaining the spatial resolution of the original image after the edge enhancement.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2006년도 Proceedings of ISRS 2006 PORSEC Volume II
• /
• pp.772-775
• /
• 2006

Muilti-dimensional feature vector matching algorithm uses multiple features such as intensity, gradient, variance, first or second derivative of a pixel to find correspondence pixels in stereo images. In this paper, we proposed a new method for adjusting automatically the weight of feature in multi-dimensional feature vector matching considering sharpeness of a pixel in feature vector distance curve. The sharpeness consists of minimum and maximum vector distances of a small window mask. In the experiment we used IKONOS satellite stereo imagery and obtained accurate matching results comparable to the manual weight-adjusting method.