A dynamic mosaicking method based on histogram equalization for an improved seamline

Abstract

Image mosaicking is a process of combining multiple images to obtain an image with a larger field of view. Errors, like ghosting and parallax effects, could appear in the mosaic if the images contain dynamic objects. This paper introduces a new idea based on histogram equalization that was used to find an optimal seamline to avoid dynamic aspects in the images. This was achieved by firstly applying a color transfer algorithm to adjust the lightness between the images in the lab color space. After that a registration was done with SIFT and RANSAC. And finally, before searching the seamline, the overall intensity of the images were fixed by changing the intensity of the underexposed and overexposed areas with a histogram equalization method. As a result, the path of the seamline was improved. The present algorithm gives a fast result and a better detection of the dynamic objects, thus a better a mosaic result.