Seamless Stitching of Images

Dr. Peter Hampton, a controls engineer here at Cellula Robotics Ltd., has recently co-authored a conference paper titled “Seamless Stitching of Images Based on a Haar Wavelet 2D Integration Method” with Ioana Sevcenco and Dr. Pan Agathoklis of the University of Victoria.

This paper is on mosaicking images in a manner that hides the borders of the sub-images.  The paper focuses on photographic images, but has potential for generating seamless maps from underwater surveys.  The process is resilient to misregistrations of adjacent images.  Further, the computation cost scales linearly with the number of pixels, allowing for much larger data sets to be processed than competing methods.

  Left : Two misregistered images with different light intensity mosaicked together.  Right:  Reconstructed image using Dr. Hampton’s patented process and the same images.

Left: Two misregistered images with different light intensity mosaicked together. Right: Reconstructed image using Dr. Hampton’s patented process and the same images.

The paper is:

I. Sevcenco, P.J. Hampton, P. Agathoklis “Seamless Stitching of Images Based on a Haar Wavelet 2D Integration Method” to be published in IEEE’s 17th Int. Conf. on Digital Signal Processing (DSP2011), July 2011