摘要
提出一种基于小波及分形理论的插值算法,用于互有位移序列图像的重构,它既考虑了各幅图像的细节特征又考虑了图像的全局特征,使得到的重构图像更接近原图。阐述了超分辨力复原技术的理论基础及复原算法,指出了基于序列图像的高分辨力成像技术的合理性,并利用小波及分形相结合的方法对互有位移的双幅图像进行了重构,实验和仿真的复原图像表明,这种方法是可行和有效的。与传统方法相比,此种方法的PSNR提高了约5dB。
An interpolation algorithm basing on wavelet and fractal theory used for image reconstruction is proposed. This approach not only keeps fine details of images but also offers the characteristics of the whole image. So it will be more approached to original image. The theories of super-resolution for imaging restoration are explained. Meanwhile the rationality of high-resolution technique basing on image sequence is introduced. The two images with displacement are reconstructed by the algorithm of wavelet with fractal. The result of experiment and computer simulations show that this algorithm is feasible and efficient. The results show that the algorithm is able to get about 5dB PSNR raise compared with traditional interpolation methods.
出处
《光学仪器》
2005年第6期23-28,共6页
Optical Instruments
关键词
小波变换
分形
超分辨力
重构
wavelet transform
fractal
super-resolution
reconstruction
