Image Mosaicing Based on Normalized Correlation and Rectangle - to - Quadrilateral Perspective Transformation 


Vol. 8,  No. 3, pp. 311-318, Jun.  2001
10.3745/KIPSTB.2001.8.3.311


PDF
  Abstract

In this paper we propose an algorithm which finds the planar perspective transformation between overlapping image pairs. It is based on the normalized correlation and the rectangle-to-quadrilateral projective projection. We calculate the global translation by using a block matching. Then to find 4 correspondence points that maximize the normalized correlation between the overlapping regions, we use SA (simulated annealing) algorithm in Gaussian image pyramid. The planer perspective transformation can be calculated by the rectangle-to-quadrilateral mapping using the correspondence points. Finally, we blend RGB colors of the overlapping region by linear weights. In experiments we show results that merge 3 images into a large mosaic image.

  Statistics


  Cite this article

[IEEE Style]

D. K. Kim and B. T. Jang, "Image Mosaicing Based on Normalized Correlation and Rectangle - to - Quadrilateral Perspective Transformation," The KIPS Transactions:PartB , vol. 8, no. 3, pp. 311-318, 2001. DOI: 10.3745/KIPSTB.2001.8.3.311.

[ACM Style]

Dong Keun Kim and Byung Tae Jang. 2001. Image Mosaicing Based on Normalized Correlation and Rectangle - to - Quadrilateral Perspective Transformation. The KIPS Transactions:PartB , 8, 3, (2001), 311-318. DOI: 10.3745/KIPSTB.2001.8.3.311.