DocumentCode :
1715466
Title :
A novel image mosaic method based on improved SIFT algorithm and Contourlet Transform
Author :
Cheng Cheng ; Ma Li-ling ; Yu Yuan-Long ; Wang Jun-zheng
Author_Institution :
Autom. Sch., Beijing Inst. of Technol., Beijing, China
fYear :
2013
Firstpage :
3657
Lastpage :
3661
Abstract :
The two key steps in image mosaic process are image registration and image fusion. The common SIFT algorithm needs complicated calculation, which wastes too much time in practical image registration. This problem is solved in this paper by giving the feature point in SIFT a more accurate main direction. Thus, the SIFT can be more easily used. Based on this improved SIFT algorithm, a new image mosaic method is proposed by combining it with the Contourlet Transform. An advantage of using Contourlet Transform in image fusion is that it can eliminate the color difference and make the merged picture clear. In this new method, the improved SIFT is used to do the registration work and the Contourlet Transform is adopted to merge the two images after registration. An experiment is also conducted in this paper to assess the stitching effect of the proposed method. The experiment result shows that this new method can accelerate the SIFT process in image registration and enhance the image mosaic effect.
Keywords :
feature extraction; image colour analysis; image fusion; image registration; image segmentation; transforms; SIFT algorithm; color difference; contourlet transform; feature point; image fusion; image mosaic method; image mosaic process; image registration; stitching effect; Irrigation; Contourlet; Image mosaic; SIFT; fusion; registration;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Control Conference (CCC), 2013 32nd Chinese
Conference_Location :
Xi´an
Type :
conf
Filename :
6640056
Link To Document :
بازگشت