• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 2006년도 춘계공동학술대회 논문집
• /
• pp.1677-1683
• /
• 2006

Color image segmentation is useful for fast retrieval in large image database. For that purpose, new image segmentation technique based on the probability of pixel distribution in the image is proposed. Color image is first divided into R, G, and B channel images. Then, pixel distribution from each of channel image is extracted to select to which it is similar among the well known probabilistic distribution function-Weibull, Exponential, Beta, Gamma, Normal, and Uniform. We use sum of least square error to measure of the quality how well an image is fitted to distribution. That P.d.f has minimum score in relation to sum of square error is chosen. Next, each image is quantized into 4 gray levels by applying thresholds to the c.d.f of the selected distribution of each channel. Finally, three quantized images are combined into one color image to obtain final segmentation result. To show the validity of the proposed method, experiments on some images are performed.

• 대한전자공학회논문지SP
• /
• 제40권2호
• /
• pp.1-10
• /
• 2003

Watershed 알고리즘은 형태학 분야에서 연구되어 온 것으로 단순화된 영상에 대한 경사 영상 화소의 밝기 값을 고도로 생각함으로써 영상을 분할하는데 많이 적용하였다. 하지만, 노이즈에 의해 훼손된 영상을 분할 할 경우, 수 많은 local minima로 인해 영상이 과 분할되고, 분할된 영역을 병합하기 위한 계산 시간 증가의 문제점이 발생된다. 이러한 문제점을 해결하기 위해, 본 논문에서는 웨이블릿 변환을 사용한 watershed 기반 영상 분할의 효율성 증대를 위한 방법을 제안한다. 제안한 영상 분할 방법은 웨이블릿 변환을 이용한 영상의 계층적 표현인 피라미드 표현 단계, watershed 알고리즘을 이용한 영상 분할 단계, 웨이블릿 계수(coefficient)를 이용한 영역 병합 단계와 웨이블릿 역 변환(inverse wavelet transform)을 이용한 영역 투영 단계고 구성된다. 제안된 방법은 노이즈가 포함된 훼손된 영상을 분할 시 발생하는 과 분할문제를 감소시킬 뿐만 아니라, 분할 성능의 개선됨을 알 수 있다.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2008년도 International Symposium on Remote Sensing
• /
• pp.72-75
• /
• 2008

Image segmentation techniques becoming increasingly important in the field of remote sensing image analysis in areas such as object oriented image classification. This paper presents a new method for image segmentation in High Resolution Remote Sensing Image based on Seeded Region Growing (SRG) and Edge Information. Firstly, multi-spectral edge detection was done using an entropy operator in pan-sharpened QuickBird imagery. Then, the initial seeds were automatically selected from the obtained edge map. After automatic selection of significant seeds, an initial segmentation was achieved by applying SRG. Finally the region merging process, using region adjacency graph (RAG), was carried out to get the final segmentation result. Experimental results demonstrated that the proposed method has good potential for application in the segmentation of high resolution satellite images.

• 전자공학회논문지CI
• /
• 제45권1호
• /
• pp.11-24
• /
• 2008

효율적인 저차원의 인덱싱을 제공하기 위해 이미지를 유사한 성질을 갖는 영역으로 나누고, 나누어진 영역에 대해 유사성을 비교하는 영역 기반 이미지 검색이 제안되었다. 그러나 영역 기반 이미지 검색은 이미지를 유사한 영역으로 나누기 위한 이미지 세그멘테이션 기술이 추가적으로 필요하다. 일반적인 칼라 자연 이미지의 경우 다양한 칼라와 질감 성분을 갖는 영역으로 나누는 것은 많은 어려움이 있다. 본 논문에서는 자동적인 칼라 이미지 세그멘테이션 알고리즘을 제안한다. 제안하는 세그멘테이션 방법은 양자화를 통해 칼라수를 줄이고 양자화 된 이미지를 Fisher의 클래스 선형 판별식을 이용하여 이미지의 전체적인 에지를 보여주는 그레이 레벨 이미지를 생성한다. 이렇게 얻은 그레이 레벨 에지 이미지를 지역적 임계치 비교를 통해 이진 에지 이미지로 변환하고 이진 에지의 끊어진 부분을 찾아내어 인접 에지에 연결하여 영역을 생성한다. 마지막으로 나누어진 영역간의 유사성을 비교하고 유사한 영역을 병합하여 최종 세그멘테이션 결과 이미지를 생성한다. 본 논문에서는 세그멘테이션 알고리즘을 이용한 영역 기반 이미지 검색 시스템을 구현하였으며, 다양한 실험에 의하면 제안한 세그멘테이션 방법이 다양한 이미지에 대하여 양질의 세그멘테이션 결과를 보이는 것으로 나타났다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2004년도 학술대회지
• /
• pp.640-644
• /
• 2004

Since the existing quadtree image segmentation methods decide the presence of image information using the maximum and minimum pixel value within an image block, they are very sensitive to noise. Although many image segmentation methods have been researched up to date, they can not execute the optimum image segmentation if noise is included in an image because there is no accurate parameters which can distinguish noise. For that reason, all application using the existing quadtree segmentation has potential of decreasing in performance due to noise. This paper proposed a quadtree image segmentation based on SDT (Standard Deviation Threshold) that can effectively extract image information parameters from a noise image. This method has the advantage of distinguishing the presence of image information even if the image has noises caused by communication. Furthermore, this paper verified through test comparison that the proposed quadtree segmentation could estimate more accurate image information parameters than the existing ones even in noise-added environment.

• 제어로봇시스템학회논문지
• /
• 제8권4호
• /
• pp.333-337
• /
• 2002

A content-based image retrieval method using color, texture, and shape features is proposed in this paper. A region segmentation technique using PIM(Picture Information Measure) entropy is used for similarity indexing. For segmentation, a color image is first transformed to a gray image and it is divided into n$\times$n non-overlapping blocks. Entropy using PIM is obtained from each block. Adequate variance to perform good segmentation of images in the database is obtained heuristically. As variance increases up to some bound, objects within the image can be easily segmented from the background. Therefore, variance is a good indication for adequate image segmentation. For high variance image, the image is segmented into two regions-high and low entropy regions. In high entropy region, hue-saturation-intensity and canny edge histograms are used for image similarity calculation. For image having lower variance is well represented by global texture information. Experiments show that the proposed method displayed similar images at the average of 4th rank for top-10 retrieval case.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2002년도 추계종합학술대회
• /
• pp.804-807
• /
• 2002

영상의 에지는 입력 영상에 대한 많은 정보들을 가지고 있다. 에지 검출을 이용하는 많은 응용들이 있으며, 다양한 특수 효과들을 위해 사용되기도 한다. 에지 검출은 영상 분석의 한 분야로서 영상분할은 영상의 구성을 결정하기 위해서 화소들을 하나의 영역으로 만들기 위해 사용된다. 본 논문에서는 영상분할을 위한 에지검출의 다양한 방법들을 통한 영상분할을 하였다. 먼저 영상의 특징을 분석하고 각 영상의 특징에 따라 에지검출의 방법을 선택적으로 채택하도록 하여 영상특징을 추출하였다. 언어의 특징을 고려하여 Java를 이용한 영상분할을 통해 효율적인 에지 검출기를 구현하였다.

• 한국멀티미디어학회논문지
• /
• 제22권9호
• /
• pp.1000-1010
• /
• 2019

Image semantic segmentation and dehazing are key tasks in the computer vision. In recent years, researches in both tasks have achieved substantial improvements in performance with the development of Convolutional Neural Network (CNN). However, most of the previous works for semantic segmentation assume the images are captured in clear weather and show degraded performance under hazy images with low contrast and faded color. Meanwhile, dehazing aims to recover clear image given observed hazy image, which is an ill-posed problem and can be alleviated with additional information about the image. In this work, we propose a deep multi-task network for simultaneous semantic segmentation and dehazing. The proposed network takes single haze image as input and predicts dense semantic segmentation map and clear image. The visual information getting refined during the dehazing process can help the recognition task of semantic segmentation. On the other hand, semantic features obtained during the semantic segmentation process can provide cues for color priors for objects, which can help dehazing process. Experimental results demonstrate the effectiveness of the proposed multi-task approach, showing improved performance compared to the separate networks.

• Journal of Information Processing Systems
• /
• 제17권5호
• /
• pp.892-904
• /
• 2021

An image semantic segmentation model is proposed based on improved ENet network in order to achieve the low accuracy of image semantic segmentation in complex environment. Firstly, this paper performs pruning and convolution optimization operations on the ENet network. That is, the network structure is reasonably adjusted for better results in image segmentation by reducing the convolution operation in the decoder and proposing the bottleneck convolution structure. Squeeze-and-excitation (SE) module is then integrated into the optimized ENet network. Small-scale targets see improvement in segmentation accuracy via automatic learning of the importance of each feature channel. Finally, the experiment was verified on the public dataset. This method outperforms the existing comparison methods in mean pixel accuracy (MPA) and mean intersection over union (MIOU) values. And in a short running time, the accuracy of the segmentation and the efficiency of the operation are guaranteed.

• 전기학회논문지P
• /
• 제51권1호
• /
• pp.39-44
• /
• 2002

In this paper, implementation mode image segmentation method for separate image is presented. The method of segmentation image in conventional method, the error are generated by the threshold values. To improve these problem for segmentation image, the calculation of weighting factor using brightness distribution by histogram of stored images are proposed. For safe image of object and laser image, the computed weighting factor is set to the threshold value. Therefore the image erosion and spread are improved, the correct and reliable informations can be measured. In this paper, the system of 3-D extracting information using the proposed algorithm can be applied to manufactory automation, building automation, security guard system, and detecting information system for all of the industry areas.