Mr Abdul Jabbar Shaikh Azad Nmu Papaer Submit

Embed Size (px)

Citation preview

  • 8/2/2019 Mr Abdul Jabbar Shaikh Azad Nmu Papaer Submit

    1/4

    Using Epipolar Geometry Applying the Bit- Plane on Images using Mat lab

    Mr. ABDUL JABBAR SHAIKH AZAD, Mr. KURESHI ANIS MAYODDIN,

    AND

    Dr. RAMESH R. MANZA.

    Department of Computer Science, P.S.G.V.P.s Mandals Arts, Commerce and Science College ShahadaAnd

    Department of Computer Science, Dr. Babasaheb Ambedkar Marathwada University, Aurangabad.

    Email [email protected] and [email protected]

    ______________________________________________________________

    Abstract:

    In this paper we show some results on Epipolar Geometry applying on various images using MATLAB. TheFundamental matrix is give to the depth information on the images we get the critical view of images that can be transfer in tothe Grayscale image in to binary image. Matlab is an ideal tool for simulating digital communication system, its gates easyscripting language and excellent data visualization capabilities. Images breaking them up in to their bit- planes images arecollecting their pixel values the grey value of each image of an 8- bit image an 8- bit binary word. The binary image will haveonly given the two values 0 and 1 and hence the image shows best clarity results is fine. The epipolar geometry is a key point incomputer vision and the fundamental matrix estimation, computer vision and the fundamental matrix estimation is the uniqueway to compute it.

    Keywords: Epipolar Geometry, Fundamental Matrix, Matlab and Robust methods etc.

    _____________________________________________________________________

    I. INTRODUCTION

    Epipolar geometry estimate the fundamentalmatrix have been proposed, which can be classified intothe linear methods and iterative methods that deal withbad point localization due to noise in imagesegmentation and robust techniques that eliminate theoutliers due to false matching. Epipolar geometryallows us to clarify what information is needed in orderto perform the search for corresponding elements onlyalong image lines.

    Several methods to estimate the fundamentalmatrix have been proposed, which can be classified intolinear methods and iterative methods that deal with badpoint localization due to noise in image segmentation,and robust techniques that eliminate the outliers due tofalse matching. Epipolar geometry allows us to clarifywhat information is needed in order to perform thesearch for corresponding elements only along imagelines. The practical importance of Epipolar geometry isbased on the fact that the Epipolar plane intersects eachimage in a line called Epipolar line. We are applyingbit plane on image and those image can be changed hisbit point with the best clear results. Grayscale imagecan be transformed into a sequence of binary images bybreaking them up into their bit-planes. If we consider

    the grey value of each image pixel on an 8-bit as an 8-bit binary word, then those the images 0th

    bit planeconsists of the last bit of each grey value. And hence theimage has least effect (least significant bit plane). The7

    thbit plane consists of the first bit in each value; it is

    called the most significant bit, and the plane consistingof those bits. If we take a gray scale image, we start bymaking it a matrix of type double, this means we canperform asthmatic on the values.

    Example:

    >>c=imread (xyz.jpg);

    >>cd =double;

    We now isolate the bit planes by simplydividing the matrix cd by successive power of 2,throwing away the remainder and seeing if the imagebit is 0 or 1. We can apply some mod function on thatimages and then getting the result of the image with thegrey scale.

    II.

    COMPARISION OF EPIPOLAR LINESEpipolar lines serves as qualitative check of the

    correctness of our result. Following some figure showsthe method of using Matlab can be comparison to theapplying with geometry of the images. The grayscaleimage will be getting most clarity of picture. We showsome example of the quality of the fundamental matrixfrom the vision of images. The epipolar lines should bepassing exactly through points corresponding to markedpoints on the left to right images. Following image isexisting original is image. And we are applying bitplane on that image using Matlab. All over this paperwe describe a new linear method for good estimatingepipolar geometry that has the pleasing stabilityproperties of the Hartley method and yet is alsoinvariant to equiform transformations of the bit planeimages. Furthermore the method can be readilyextended to other some images quantities such ashomographies.

    The epipolar geometry between two views isessentially the geometry of the intersection of the imageplanes with the pencil of planes having the baseline asaxis (the baseline is the line joining the camera centers).This geometry is usually motivated by considering thesearch for corresponding points in stereo matching, and

  • 8/2/2019 Mr Abdul Jabbar Shaikh Azad Nmu Papaer Submit

    2/4

    we will start from that objective here. Suppose a pointX in 3-space is imaged in two views, at x in the first,and x in the second. What is the relation between thecorresponding image points x and x as shown in figure8.1 the image points x and x, space point X, and

    camera centers are coplanar. Denote this plane as

    clearly, the rays back-projected from x and xintersect

    at X, and the rays are coplanar, lying. It is this latterproperty that is of most significance in searching for acorrespondence. Supposing now that we know only x,we may ask how the corresponding point X isconstrained. The plane is determined by the baselineand the ray defined by x. From above we know that theray corresponding to the (unknown) point x lies in ,hence the point x lies on the line of intersection lofwith the second image plane. This line lis the image inthe second view of the ray back-projected from x. Interms of a stereo correspondence algorithm the benefitis that the search for the point corresponding to x neednot cover the entire image plane but can be restricted tothe line l.

    Figure (A): Points Correspondence geometry: (a)the two cameras are indicated by their center C and Cand image planes. The camera canters space point x,and its image x and x lie in common plane . (b) animage point x a ray 3-defined by first camera center C

    and x. This ray is imaged as line l in the second view.The 3 space point X which projects to x must lie on thisray, so the image of X in the second view must lie on l.

    Figure1. Original Image

    The above image is the colorful then the clarityof points matching is poor. And hence we areconverting the image into the grayscale image, becausematching points accuracy is best. We use thefundamental matrix F fits the putative correspondenceset X. The estimation of the images we are applyingrotation and translation. We compute this fit measure

    using a robust combination of the images which isapproximation of the geometric re- projection error.Some are the individual error of the putativecorrespondence both the inliers and outliers need to becombined into one overall image results as follows.

    Figure2. Applying Bit Plane on original image >>

    c0=mod (cd, 2);

    Figure3. Applying Bit Plane on original image >>c1=mod (floor (cd/2), 2);

    Figure4. Applying Bit Plane on original image >>c2=mod (floor (cd/4), 2);

    For the applying bit plane on those images theresult are coming extremely clear hence we areconverting those images into the grayscale image. Theimage represents an image as a matrix where everyelement has a value corresponding to how bright anddark at the corresponding position should be colored.There are two ways to represent the number that re-present the brightness of the pixel.

  • 8/2/2019 Mr Abdul Jabbar Shaikh Azad Nmu Papaer Submit

    3/4

    Figure5. Applying Bit Plane on original image >>c3=mod (floor (cd/8), 2);

    Figure6. Applying Bit Plane on original image >>c4=mod (floor (cd/16), 2);

    Figure7. Applying Bit Plane on original image >>

    c5=mod (floor (cd/32), 2);

    Figure8. Applying Bit Plane on

    original image >> c6=mod (floor (cd/64), 2);

    Figure9. Applying Bit Plane on

    original image >> c7=mod (floor (cd/128), 2);

    On the left bottom window of fig. 9 we can

    observe the feature points (crosses) moving from initial

    towards the bit plane image. We are considering two

    possible approaches to solve this problem. We would be

    able to applied directly onto the entities facades

    assumed to be planner or more complex approaches to

    solve this problem. For all observation to the applyingbit plane on to the images we are get more accuracy for

    the matching point on to the images. There for exactly

    matching points are not yet considering we are going to

    automat matching points on various images.

    III. CONCLUSION AND FUTUREWORK

    The System represent in this paper describesthe images reconstruct process which takes only a pairof aerial photos and described system does not requireprior knowledge of the real images is the necessarywork to design three dimensional models is

    substantially reduced. Regarding the future work, ourobjective is to obtain deep information about extractedfrom the video streams captured in the bus, vehicles etc.epipolar geometry principle that uses the main detectionmatching points on those images.

    IV. REFERENCES

    [1] S. Agarwal, N. Snavely, I. Simon, S. Seitz, and R.Szeliski. Building rome in a day. In Proceedings of ICCV,pages 7279, Kyoto, Japan, 2009.

    [2] Y. Avrithis, Y. Kalantidis, G. Tolias, and E. Spyrou.Retrieving landmark and non-landmark images fromcommunity photo collections. In Proceedings ofACMMultimedia, pages 153162, Firenze, Italy, October 2010.

    [3] N. Snavely, S. Seitz, and R. Szeliski. Photo tourism:Exploring image collections in 3d. In SIGGRAPH06,pages 835846, Boston, Massachusetts, USA, 2006.

    [4] B. Triggs. Joint feature distributions for imagecorrespondence. In Proceedings of ICCV,pages 201208,Vancouver, British Columbia, Canada, 2001.

    [5] P. A. Beardsley, P. H. S. Torr, and A. Zisserman. 3Dmodel acquisition from extended image sequences. InProceedings of the 4th European Conference on ComputerVision, LNCS 1065, Cambridge, pages 683695, 1996.

    [6] P. A. Beardsley, A. Zisserman, and D. W. Murray.Sequential update of projective and affine structure from

  • 8/2/2019 Mr Abdul Jabbar Shaikh Azad Nmu Papaer Submit

    4/4

    motion. International Journal of ComputerVision,23(3):235259, 1997.

    [7] F. Bookstein. Fitting conic sections to scattered data.Computer Vision, Graphics and Image Processing, 9:5671, 1979.

    [8] M. Brooks, L. De Agapito, D. Huynh, and L. Baumela.Towards robust metric reconstruction via a dynamicuncalibrated stereo head. Image and Vision Computing,16(14):9891002, 1998.

    [9]

    O. D. Faugeras and M. Hebert. The representation,recognition, and locating of 3-D objects. InternationalJournal of Robotics Research, 5(3):2752, 1992.

    [10] W.Zhang and J. Kosecka Image based Localization inurban environments. In proceeding of 3DPVT, pages 33-40, Washington DC USA 2006.

    [11] J.Piazzi and D.Prattichizzo. An auto-epipolar strategy formobile robot Visual Servoing. In IEEE IROS03 Conf. on

    Intelligient Robots and Systems, Las Vegas, USA , 2003.

    [12] Bastos, R., Dias, J. M. S., Fully Automated TextureTracking Based on Natural Features Extraction andTemplate Matching, submitted to ISMAR 04, 3rd IEEEand ACM International Symposium on Mixed andAugmented Reality, in Arlington, VA, USA, Nov. 2-5,2004.

    [13] Nakagawa, M., Shibasaki R., Kagawa Y., Fusing stereolinear CCD image and laser range data for building 3Durban model, ISPRS Commisssion IV, WG IV/7, 2002.

    [14] Mr. Abdul Jabbar Shaikh Azad and Dr. Ramesh R. Manzafrom Epipolar Geometry for Stereo Image Processing forgood Estimation of the Fundamental Matrix. 2011.

    [15] A. Wahle, H. Oswald, and E. Fleck, "3D heart-vesselreconstruction from biplane angiograms,"IEEE ComputerGraphics and Applications, vol. 16, pp. 65-73, 1996.

    [16] A. M. Lopez, D. Lloret, J. Serrat, and J. J. Villanueva,"Multilocal creaseness based on the level-set extrinsiccurvature," Computer Vision and Image Understanding,vol. 77, pp. 111-144, 2000.

    [17] A. Daugherty, "Mouse models of atherosclerosis,"American Journal of the Medical Sciences, vol. 323, pp. 3-10, 2002.

    [18] A. Daugherty and D. L. Rateri, "Development ofexperimental designs for atherosclerosis studies in mice,"

    Methods, vol. 36, pp. 129-138, 2005.[19] M. Narayanaswamy, K. C. Wright, and K. Kandarpa,

    "Animal models for atherosclerosis, restenosis, andendovascular graft research," Journal of Vascular and

    Interventional Radiology , vol. 11, pp. 5-17, 2000.

    [20] S. Aylward and E. Bullitt, Initialization, Noise,Singularities and Scale in Height Ridge Traversal forTubular Object Centerline Extraction, IEEE Trans.Medical Imaging, vol. 21, no. 2, 2002, pp. 61-75.

    [21] B. Jolly et al., Needle Tracking and Detection in the TIPSEndovascular Procedure, Proc. Medical ImageComputing and Computer-Assisted Intervention (MICCAI2003), LNCS 2878, Springer-Verlag, 2003, pp. 953-954.

    [22] B. Girod, and K. W. Stuhlm uller. A Content-DependentFast DCT for Low Bit-Rate Video Coding, in Proc. IEEEInt. Conf. Image Process., Chicago, Illinois, Oct. 1998,vol. 3, pp. 80-83.

    [23]

    X. Dong, Y. Pang, J. Wen, G. Wang, W. Li, Y. Gao, andS. Yang. A Fast Efficient Algorithm for Enhancement ofLow Lighting Video, in Int. Conf. Digital Media andDigital Content Management, Chongqing, China, Dec.2010.

    [24] J. Barreto, F. Martin, and R. Horaud, Visualservoing/tracking using central catadioptric images, in

    Int. Symposium on Experimental Robotics, Ischia, Italy,Jul. 2002.