26 이경승(394~400).hwp

Size: px

Start display at page:

Download "26 이경승(394~400).hwp"

철 금
7 years ago
Views:

1 위치와색상정보를사용한 SURF 정합성능향상기법 이경승 *, 김대훈 *, 노승민 **, 황인준 * KyungSeung Lee *, Daehoon Kim *, Seungmin Rho ** and Eenjun Hwang * 요약 SURF(Speeded Up Robust Features). SIFT(Scale Invariant Feature Transform).,.,,. SURF. orientation. color histogram.. Abstract SURF is a robust local invariant feature descriptor and has been used in many applications such as object recognition. Even though this algorithm has similar matching accuracy compared to the SIFT, which is another popular feature extraction algorithm, it has advantage in matching time. However, these descriptors do not consider relative location information of extracted interesting points to guarantee rotation invariance. Also, since they use gray image of original color image, they do not use the color information of images, either. In this paper, we propose a method for improving matching performance of SURF descriptor using the color and relative location information of interest points. The location information is built from the angles between the line connecting the centers of interest points and the orientation line constructed for the center of each interest points. For the color information, color histogram is constructed for the region of each interest point. We show the performance of our scheme through experiments. Key words : SURF, image matching, color feature, location feature I. 서론 PC., GPS,,. * (School of Electrical Engineering, Korea University) ** (Division of Information and Communication, Baekseok University) 1 (First Author) :, (Corresponding Author) : :

2 SURF ;,,, 395 PC. SURF, [1][2], [3], [4][5]. SIFT[6][7], PCA-SIFT[8], GLOH[9], LESH[10], SURF[11]. SURF. SURF. SURF.. 1.(a).. 1.(b),. SURF. 1.(c) SURF.. SURF.. 2 SURF, 3. 4, 5. Ⅱ. 관련연구. SURF,, SURF. (a) 정상적인정합 (b) 상대적인위치가어긋난정합 (c) 동일한외형에다른색을가진객체 그림 1. 기존 SURF 기술자의문제점 Fig. 1 Problem of exsiting SURF descriptor 2-1 특징점검출 [12],. (.1), NMS.. (1)

3 그림 2. 알고리즘구조도 Fig 2. Structure of algorithm (Fast Hessian Detector) x, y, xy. 2-2 기술자추출. (Orientation) 6σ Haar. π/3 Orientation Window 360 Window,. x, y Harr 64. SURF Haar 개선된 SURF에관한다른연구 최근모바일단말기에탑재될수있는객체인식방법으로비교적빠르고정확한 SURF 알고리즘이주목을받으면서 SURF의성능개선에관한많은연구가수행되었다. 특히 SURF는회색조영상에서수행되므로색상정보가무시되는데, [13] 에서는이런점에착안하여이미지에서가우시안색상모델 (Gaussian Color Model) 을기반으로생성한 CISURF(Color Invariance based SURF) 라는기술자를사용하여성능을향상시켰다. [14] 에서는지역색상커널히스토그램 (Local Color Kernel Histogram) 을기존 SURF 기술자벡터에추가하여객체인식에사용하였다. [15] 에서는색상정보와 SURF의지역특징영역보다 10배증가시켜헤시안검출기로전역벡터 (Global Vector) G가정합에사용되었다. 본연구에서는색상정보만이아니라, 다른연구에서다루지않았던특징점의상대적인위치정보까지활용하여정합성능을높인다. Ⅲ. 색상정보와위치정보를포함하는 SURF Orientation SURF. 2. SURF

SURF ;,,, 397... 3-1 색상정보를이용한 SURF SURF,. SURF RGB,. SURF RGB.. 3-2 위치정보를이용한 SURF URF SURF. SURF. SURF.. SURF. 그림 3. RGB 값추출 Fig 3.

4 SURF ;,,, 색상정보를이용한 SURF SURF,. SURF RGB,. SURF RGB 위치정보를이용한 SURF URF SURF. SURF. SURF.. SURF. 그림 3. RGB 값추출 Fig 3. Extraction of RGB value 3 SURF. 6σ. 256 RGB RGB 16, RGB 48. SURF 그림 4. 주방향과특징점중심이이루는각 Fig 4. Angle between orientation and center of interest point

398 16 2 2012 4. 4. (a) SURF. (b). (c). (d). SURF,. 10.. 10 3. 90 42 48. 90 42. Ⅳ. 실험. Intel Core 2 Duo 2.67Ghz, 4GB Windows 7. MATLAB, OpenCV SURF mex. ALOI[16]. 1000,,,.

5 (a) SURF. (b). (c). (d). SURF, Ⅳ. 실험. Intel Core 2 Duo 2.67Ghz, 4GB Windows 7. MATLAB, OpenCV SURF mex. ALOI[16]. 1000,,,. 5 DB. 72., 그림 5. 객체인식률실험결과 Fig 5. Experiment result of object recognition. SURF 70%. true positive true negative., SURF false alarm. SURF SURF. SURF SURF. Ⅴ. 결론 SURF.

6 SURF ;,,, 399 color histogram.. SURF. 감사의글 2011 ( ) ( ) IT (NIPA-2012-H ) 참고문헌 [1] P. Schugerl, R. Sorchag, W. Bailer, and G.Thallinge r, Object re-detection using sift and mpeg-7 color descriptors, In International Workshop on Multime dia Content Analysis and Mining, pp , 2007 [2] Zhang, W., Yu, B., Zelinsky, G. J., & Samaras, D., Object class recognition using multiple layer boost ing with heterogeneous features, In CVPR 05, pp , 2005 [3] Cheng Hongwei, Jiang Letian, Chen Xiaonming, Ba o Jing, Intergrated Color Balance Algorithm For Panoramic Image, American Journal of Engineerin g and Technology Research [4] M. Brown, D. G. Lowe, Unsupervised 3D Object Recognition and Reconstruction in Unordered Data sets, Proceeding of the Fifth International Confere nce on 3-D Digital Imaging and Modeling, pp , June 13-16, [5] Liu, J., Hubbold, R., Automatic camera calibration and scene reconstruction with scale-invariant featur es, In:ISVC, pp , 2006 [6] D.G.Lowe, Object Recognition from Local Scale-I nvariant Features, Proc. Seventh Int l Conf. Comp uter Vision, (1999) [7] D.G.Lowe, Distinctive Image Features from Scale-I nvariant Keypoints, Int l J. Computer Vision, 60 (2), (2004) [8] Y. Ke, and R. Sukthankar, PCA-SIFT: A More Dist inctive Representation for Local Image Descriptors, Proc, Computer Vision and Pattern Recognition, (2004) [9] Mikolajczyk, and K., S, chmid, C, A performance evaluation of local descriptors, PAMI 27(10), (2005) [10] Sarfraz S., and Hellwich O, Head Pose Estimation in Face recognition across Pose Scenarios, Proceed ing of VISAPP 2008, Int. conference on Computer Vision theory and Applications, Madeira, Portugal, (2008) [11] H. Bay, T. Tuytelaars, and L. V. Gool, SURF: Speeded up robust features, European Conference on Computer Vision, (2006) [12] P.A. Viola, and M.J.Jones, Rapid object detection using a boosted cascade of simple features, CVPR, (1), (2008) [13] Gang Meng, Zhiguo Jiang, Danpei Zhao, The Usa ge of Color Invariance in SURF Proc. of SPIE(200 9). [14] Pen Fan, Aidong Men, Mengyang Chen, Bo Yang, Color-Surf: A Surf Descriptor with Local Kernel Color Histograms, Proceeding of IC-NIDC, pp (2009) [15] Hyunsup Yoon, Hwan-Ik Chung, Hernsoo Hahn, SURF Algorithm with Color and Global Character istics Proceeding of ICCAS-SICE, pp (20 09) [16] ALOI(Amsterdam Library of Object Images), /staff.science.uva.nl/~aloi/

400 16 2 2012 4 이경승 ( 李暻承 ) 김대훈 ( 金大勳 ) 2010 : ( ) 2010 ~

( ) 2003 : 2008 : 2008 ~ 2009 : Carnegie Mellon

Recommendation, Affective Computing, Swarm Intelligence

of Maryland at College Park 1998 ~ 1999 : Bowie State Univ.

7 이경승 ( 李暻承 ) 김대훈 ( 金大勳 ) 2010 : ( ) 2010 ~ : :,,, 2006 : ( ) 2008 : 2008 ~ : :,,, 노승민 ( 盧承民 ) 2001 : ( ) 2003 : 2008 : 2008 ~ 2009 : Carnegie Mellon University, 2009 ~ 2011 :, 2012 ~ :, : Music Retrieval and Recommendation, Affective Computing, Swarm Intelligence 황인준 ( 黃仁俊 ) 1988 : ( ) 1990 : 1998 : Univ. of Maryland at College Park 1998 ~ 1999 : Bowie State Univ., Assistant Professor 1999 ~ 1999 : Hughes Research Lab ~ 2004 : 2004 ~ : :,,,,,

High Resolution Disparity Map Generation Using TOF Depth Camera In this paper, we propose a high-resolution disparity map generation method using a lo

High Resolution Disparity Map Generation Using TOF Depth Camera In this paper, we propose a high-resolution disparity map generation method using a low-resolution Time-Of- Flight (TOF) depth camera and