Author(s): Jignesh Sarvaiya, Suprava Patnaik, Hemant Goklani
Journal: International Journal of Image Processing
ISSN 1985-2304
Volume: 4;
Issue: 2;
Start page: 119;
Date: 2010;
VIEW PDF
DOWNLOAD PDF 
Original page
Keywords: Image Registration | NSCT | Contourlet Transform | Zernike Moment.
ABSTRACT
Image registration is a process of matching images, which are taken at differenttimes, from different sensors or from different view points. It is an important stepfor a great variety of applications such as computer vision, stereo navigation,medical image analysis, pattern recognition and watermarking applications. Inthis paper an improved feature point selection and matching technique for imageregistration is proposed. This technique is based on the ability of NonsubsampledContourlet Transform (NSCT) to extract significant features irrespective offeature orientation. Then the correspondence between the extracted featurepoints of reference image and sensed image is achieved using Zernike moments.Feature point pairs are used for estimating the transformation parametersmapping the sensed image to the reference image. Experimental results illustratethe registration accuracy over a wide range for panning and zooming movementand also the robustness of the proposed algorithm to noise. Apart from imageregistration proposed method can be used for shape matching and objectclassification.
Journal: International Journal of Image Processing
ISSN 1985-2304
Volume: 4;
Issue: 2;
Start page: 119;
Date: 2010;
VIEW PDF
DOWNLOAD PDF Original pageKeywords: Image Registration | NSCT | Contourlet Transform | Zernike Moment.
ABSTRACT
Image registration is a process of matching images, which are taken at differenttimes, from different sensors or from different view points. It is an important stepfor a great variety of applications such as computer vision, stereo navigation,medical image analysis, pattern recognition and watermarking applications. Inthis paper an improved feature point selection and matching technique for imageregistration is proposed. This technique is based on the ability of NonsubsampledContourlet Transform (NSCT) to extract significant features irrespective offeature orientation. Then the correspondence between the extracted featurepoints of reference image and sensed image is achieved using Zernike moments.Feature point pairs are used for estimating the transformation parametersmapping the sensed image to the reference image. Experimental results illustratethe registration accuracy over a wide range for panning and zooming movementand also the robustness of the proposed algorithm to noise. Apart from imageregistration proposed method can be used for shape matching and objectclassification.
