• Title/Summary/Keyword: Fourier image

Search Result 415, Processing Time 0.024 seconds

A Study on the Optimization of Convolution Operation Speed through FFT Algorithm (FFT 적용을 통한 Convolution 연산속도 향상에 관한 연구)

  • Lim, Su-Chang;Kim, Jong-Chan
    • Journal of Korea Multimedia Society
    • /
    • v.24 no.11
    • /
    • pp.1552-1559
    • /
    • 2021
  • Convolution neural networks (CNNs) show notable performance in image processing and are used as representative core models. CNNs extract and learn features from large amounts of train dataset. In general, it has a structure in which a convolution layer and a fully connected layer are stacked. The core of CNN is the convolution layer. The size of the kernel used for feature extraction and the number that affect the depth of the feature map determine the amount of weight parameters of the CNN that can be learned. These parameters are the main causes of increasing the computational complexity and memory usage of the entire neural network. The most computationally expensive components in CNNs are fully connected and spatial convolution computations. In this paper, we propose a Fourier Convolution Neural Network that performs the operation of the convolution layer in the Fourier domain. We work on modifying and improving the amount of computation by applying the fast fourier transform method. Using the MNIST dataset, the performance was similar to that of the general CNN in terms of accuracy. In terms of operation speed, 7.2% faster operation speed was achieved. An average of 19% faster speed was achieved in experiments using 1024x1024 images and various sizes of kernels.

Image Clustering using Color, Texture and Shape Features

  • Sleit, Azzam;Abu Dalhoum, Abdel Llatif;Qatawneh, Mohammad;Al-Sharief, Maryam;Al-Jabaly, Rawa'a;Karajeh, Ola
    • KSII Transactions on Internet and Information Systems (TIIS)
    • /
    • v.5 no.1
    • /
    • pp.211-227
    • /
    • 2011
  • Content Based Image Retrieval (CBIR) is an approach for retrieving similar images from an image database based on automatically-derived image features. The quality of a retrieval system depends on the features used to describe image content. In this paper, we propose an image clustering system that takes a database of images as input and clusters them using k-means clustering algorithm taking into consideration color, texture and shape features. Experimental results show that the combination of the three features brings about higher values of accuracy and precision.

Showing Morphological Evolution of the Strain Response Envelope of Clay with Fourier Descriptor Analysis (퓨리에 기술자를 이용한 점성토의 변형률 응답 곡선의 형상 변이 분석)

  • Kim, Taesik;Jung, Young-Hoon
    • Journal of the Korean GEO-environmental Society
    • /
    • v.18 no.3
    • /
    • pp.25-30
    • /
    • 2017
  • This paper introduces a novel method to quantify the morphological evolution of the strain response envelope. The strain response envelope is defined as an image in strain increment space corresponding to the unit stress input in stress space. Based on the shape of strain response envelopes, the deformation characteristics of soils can be described using the framework of elastic-plastic theory. Fourier descriptor analysis was used to investigate the morphological characteristics of strain response envelopes. The numerical results show that when the stress input remains in the initial yield surface the Fourier descriptors remain constant. Once the stress input crosses the initial yield surface, every descriptors deals in this study change. Numerical and experimental results of this study show that clear yielding response is only found in natural block samples. Among the Fourier descriptors, the descriptor called as asymmetry is the best for detecting the yield and is minimally sensitive to the number of input stress paths.

Soil Particle Shape Analysis Using Fourier Descriptor Analysis (퓨리에 기술자 분석을 이용한 단일 흙 입자의 형상 분석)

  • Koo, Bonwhee;Kim, Taesik
    • Journal of the Korean GEO-environmental Society
    • /
    • v.17 no.3
    • /
    • pp.21-26
    • /
    • 2016
  • Soil particle shape analysis was conducted with sands from Jumujun, Korea and Ras Al Khair, Saudi Arabia. Two hundred times enlarged digital images of the particles of those two sands were obtained with an optical microscope. The resolution of the digital images was $640{\times}320$. By conducting digital image processing, the coordinates of the soil particle boundary were extracted. After mapping those coordinates to the complex space, Fourier transformation was performed and the coefficients of each trigonometry term were computed. The coefficients reflect the shape characteristics of the sand grains and are invariant to translation. To evaluate the shape itself excluding the size of the soil particle, the coefficient was normalized by the equivalent radius of soil particle; this is called Fourier descriptor. After analyzing the Fourier descriptors, it was found that the major characteristics of Jumunjin and Ras Al Khair sands were elongation and asymmetry. Furthermore, it was found that the particle shapes reflect the self-similar, fractal nature of the textural features. The effects of resolution on soil particle shape analysis was also studied. Regarding this, it was found that the significant Fourier descriptors were not significantly affected by the image resolution investigated in this study, but the descriptors associated with textural features were affected.

Extracting Shadow area and recovering of image (영상의 그림자 영역 경계 검출 및 복원 연구)

  • Choi, Yun-Woong;Jeon, Jae-Yong;Park, Jung-Nam;Cho, Gi-Sung
    • Proceedings of the Korean Society of Surveying, Geodesy, Photogrammetry, and Cartography Conference
    • /
    • 2007.04a
    • /
    • pp.169-173
    • /
    • 2007
  • Nowadays the aerial photos is using to get the information around our spatial environment and it increases by geometric progression in many fields. The aerial photos need in a simple object such as cartography and ground covey classification and also in a social objects such as the city plan, environment, disaster, transportation etc. However, the shadow, which includes when taking the aerial photos, makes a trouble to interpret the ground information, and also users, who need the photos in their field tasks, have restriction. This study, for removing the shadow, uses the single image and the image without the source of image and taking situation. Also, this study present clustering algorism based on HIS color model that use Hue, Saturation and Intensity, especially this study used I(intensity) to extract shadow area from image. And finally by filtering in Fourier frequency domain creates the intrinsic image which recovers the 3-D color information and removes the shadow.

  • PDF

Optical Image Switching System based on BPEJTC (BPEJTC를 이용한 광 영상 스위칭 시스템)

  • 이상이;이승현;양훈기;김은수
    • Journal of the Korean Institute of Telematics and Electronics A
    • /
    • v.32A no.10
    • /
    • pp.51-63
    • /
    • 1995
  • In this paper, a new real-time optical image switching system based on the phase-typed BPEJTC is suggested. The phase filter mask which has the arbitrary position mapping function between input and output planes is constructed by using the modified JTPS of the BPEJTC. Then, the input image is convolved with this phase filter mask in the spatial frequency domain and through further Fourier transform the input image is switched to the new positions in the output correlation plane where the correlation peaks are occurred. And, based on the computer simulation results, the practical optical switched to the new positions in the output correlation plane where the correlation peaks are occurred. And, based on the computer simulation results, the practical optical switching system is opto-digitally constructed and through some experiments on image switching the possiblity of real-time implementation of the multiple optical image switching system by using the BPEJTC is suggeste.

  • PDF

A Study on the Estimation of Ocean Surface Wave Information from Marine Radar Signals (선박 레이더 영상신호를 이용한 파랑정보 검출에 관한 연구)

  • Song, Chae-Uk;Kim, Chang-Je;Moon, Seong-Bae
    • Journal of Navigation and Port Research
    • /
    • v.27 no.5
    • /
    • pp.499-504
    • /
    • 2003
  • This paper describes the system for evaluating the sea wave informations such as wave direction and wave length in real time, by using image data obtained from the marine X-band radar. We proposed here a method for automatic selection of the partial image data without the user's individual selection at the radar. We also discussed that the wave direction could be obtained by a 2-dimensional discrete Fourier transform algorithm. We carried some evaluation works on the algorithm through computer simulation. The obtained thirteen radar image data under several sea surface conditions were analyzed by the method described and the result was presented.

Inspection for Inner Wall Surface of Communication Conduits by Laser Projection Image Analysis (레이저 투영 영상 분석에 의한 통신 관로 내벽 검사 기법)

  • Lee Dae-Ho
    • Journal of Korea Multimedia Society
    • /
    • v.9 no.9
    • /
    • pp.1131-1138
    • /
    • 2006
  • This paper proposes a novel method for grading of underground communication conduits by laser projection image analysis. The equipment thrust into conduit consists of a laser diode, a light emitting diode and a camera, the laser diode is utilized for generating projection image onto pipe wall, the light emitting diode for lighting environment and the image of conduit is acquired by the camera. In order to segment profile region, we used a novel color difference model and multiple thresholds method. The shape of profile ring is represented as a minimum diameter and the Fourier descriptor, and then the pipe status is graded by the rule-based method. Both local and global features of the segmented ring shaped, the minimum diameter and the Fourier descriptor, are utilized, therefore injured and distorted pipes can be correctly graded. From the experimental results, the classification is measured with accuracy such that false alarms are less than 2% under the various conditions.

  • PDF

A Study for Introducing a Method of Detecting and Recovering the Shadow Edge from Aerial Photos (항공영상에서 그림자 경계 탐색 및 복원 기법 연구)

  • Jung, Yong-Ju;Jang, Young-Woon;Choi, Yun-Woong;Cho, Gi-Sung
    • Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
    • /
    • v.24 no.4
    • /
    • pp.327-334
    • /
    • 2006
  • The aerial photos need in a simple object such as cartography and ground cover classification and also in a social objects such as the city plan, environment, disaster, transportation etc. However, the shadow, which includes when taking the aerial photos, makes a trouble to interpret the ground information, and also users, who need the photos in their field tasks, have a restriction. Generally the shadow occurs by the building and surface topography, and the detail cause is by changing of the illumination in an area. For removing the shadow this study uses the single image and processes the image without the source of image and taking situation. Also, applying the entropy minimization method it generates the 1-D gray-scale invariant image for creating the shadow edge mask and using the Canny edge detection creates the shadow edge mask, and finally by filtering in Fourier frequency domain creates the intrinsic image which recovers the 3-D color information and removes the shadow.

Facial Image Detection with Matched Filters (Matched Filter를 이용한 얼굴 영상 검출)

  • 황인택;신명숙;최광남
    • Proceedings of the Korea Multimedia Society Conference
    • /
    • 2003.11a
    • /
    • pp.208-211
    • /
    • 2003
  • 이 논문은 Matched Filter 기술을 사용해 다양한 영상 중 얼굴 영상을 분류하는 연구에 대해서 기술한다. 기본 목표는 얼굴 영상과 그 외의 영상을 분류할 수 있는 효과적인 필터를 개발하는 것이다. 이런 Matched Filter는 Fourier 역변환을 사용해 훈련영상(Training Image)으로부터 얻을 수 있다. 실험평가는 Yale대학의 얼굴 데이터베이스의 얼굴영상과 다양한 형태를 보이는 임의의 영상으로 평가한다. 우리는 여기서 얼굴 영상을 분류하기 위한 방법 중 한가지로서 Matched Filter를 이용할 수 있음을 확인할 수 있다.

  • PDF