全景图拼接中图像融合算法的研究

Study on Image Fusion Algorithm of Panoramic Image Stitching

为了实现平滑连续的全景图拼接融合效果,针对全景图拼接的特点,提出了利用彩色空间变换和Contourlet变换结合对比度金字塔分解的图像融合算法。首先进行HSI彩色空间变换,得到图像的亮度信息,然后利用基于对比度金字塔的Contourlet变换对亮度信息进行塔式分解,得到不同的频带信息,再对不同频带进行融合处理。实验证明,该文提出的算法充分利用Contourlet变换的轮廓特性,展示了图像的细节信息,从而很好地实现了全景图的拼接融合效果。

In order to achieve smooth and continuous panorama stitching effect, an image fusion algorithm which based on contrast pyramid and combines color space conversion and Contourlet transform is proposed by considering the characteristics of panorama stitching. Firstly, luminance information of images is calculated using HSI transform. Then contrasting pyramid based Contourlet transform is used to decompose luminance information to get sub-band information of images. Finally, images are reconstructed by fusing different sub-band’s information. Experimental results show that the proposed algorithm which leverages the contour features of Contourlet transform and the detail information of images could achieve good effects on panorama stitching fusion.