Robust Image Dehazing and Matching Based on Koschmieder s Law And SIFT Descriptor

  Afthab Baik K.A,  Beena M.V

If two images of the same scene are taken into consideration, then the matching between them is a challenging task. The images which are used for the process of matching undergo geometric and photometric variations. These variations are due to the difference in time at which the images are taken, the difference in camera from which the images are taken and the difference in viewpoints. The matching of outdoor and aerial images is more challenging. Besides geometric and photometric variations, degradation by haze can also be found in outdoor and aerial images. Such degraded images cannot be used to perform robust image matching. Therefore, dehazing of hazy images is necessary. Here a contrast based dehazing technique which is based on koschmieder's law is used to effectively dehaze hazy pair of images. Koschmieder's law based dehazing technique is compared with a dehazing technique which is based on dark channel prior. Koschmieder's law based dehazing technique outperforms dark channel based dehazing technique in terms of speed. Moreover, it is able to produce a visually pleasing result. It is shown that koschmieder's law based dehazing technique is best suited for image matching since it is able to produce more number of feature points for matching than the feature points in dark channel based dehazing technique. In order to detect and describe the feature points for matching, a powerful detector and descriptor called SIFT(Scale Invariant Feature Transform) is used.

Keywords- Dehazing, image matching, Scale Invariant Feature Transform(SIFT)

Unique Identification Number - IJEDR1503126
Page Number(s) - 1-7
Pubished in - Volume 3 | Issue 3 | September 2015
DOI (Digital Object Identifier) -   
Publisher - IJEDR (ISSN - 2321-9939)

  Afthab Baik K.A,  Beena M.V,   "Robust Image Dehazing and Matching Based on Koschmieder s Law And SIFT Descriptor", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.3, Issue 3, pp.1-7, September 2015, Available at :http://www.ijedr.org/papers/IJEDR1503126.pdf