Title :
A Novel Geometric Algorithm for Blind Image Restoration Based on High-Dimensional Space
Author :
Jiang, Xuena ; Liu, Yangyang ; Wang, Shoujue
Author_Institution :
Lab. of Artificial Neural Networks, Chinese Acad. of Sci., Beijing, China
Abstract :
A novel geometric algorithm for blind image restoration is proposed in this paper, based on high-dimensional space geometrical informatics (HDSGI) theory. In this algorithm every image is considered as a point, and the location relationship of the points in high-dimensional space, i.e. the intrinsic relationship of images is analyzed. Then geometric technique of "blurring-blurring-deblurring" is adopted to get the deblurring images. Comparing with other existing algorithms like Wiener filter, super resolution image restoration etc., the experimental results show that the proposed algorithm could not only obtain better details of images but also reduces the computational complexity with less computing time. The novel algorithm probably shows a new direction for blind image restoration with promising perspective of applications.
Keywords :
Wiener filters; image resolution; image restoration; Wiener filter; blind image restoration; geometric algorithm; high-dimensional space geometrical informatics; image deblurring; image restoration; image super resolution; Algorithm design and analysis; Artificial neural networks; Image analysis; Image processing; Image resolution; Image restoration; Informatics; Information geometry; Laboratories; Wiener filter;
Conference_Titel :
Image and Signal Processing, 2009. CISP '09. 2nd International Congress on
Conference_Location :
Tianjin
Print_ISBN :
978-1-4244-4129-7
Electronic_ISBN :
978-1-4244-4131-0
DOI :
10.1109/CISP.2009.5304026