高级检索
武昆, 李桂菊, 韩广良, 杨航, 王宇庆. 四元数引导滤波彩色图像细节增强[J]. 计算机辅助设计与图形学学报, 2017, 29(3): 419-427.
引用本文: 武昆, 李桂菊, 韩广良, 杨航, 王宇庆. 四元数引导滤波彩色图像细节增强[J]. 计算机辅助设计与图形学学报, 2017, 29(3): 419-427.
Wu Kun, Li Guiju, Han Guangliang, Yang Hang, Wang Yuqing. Color Image Detail Enhancement Based on Quaternion Guided Filter[J]. Journal of Computer-Aided Design & Computer Graphics, 2017, 29(3): 419-427.
Citation: Wu Kun, Li Guiju, Han Guangliang, Yang Hang, Wang Yuqing. Color Image Detail Enhancement Based on Quaternion Guided Filter[J]. Journal of Computer-Aided Design & Computer Graphics, 2017, 29(3): 419-427.

四元数引导滤波彩色图像细节增强

Color Image Detail Enhancement Based on Quaternion Guided Filter

  • 摘要: 为了使彩色图像细节突出,可视性强,提出一种基于四元数引导滤波的彩色图像增强方法.首先将传统引导滤波推广到四元数域中,推导得出基于四元数的引导滤波;其次将彩色图像以四元数矩阵的形式建模,通过四元数引导滤波对图像的四元数矩阵进行滤波处理;再采用图像分层方法将待增强彩色图像分成一个平滑图像和一个细节图像,并对细节图像进行自适应增强处理;最后将平滑图像和增强后的细节图像重构,得到最终的增强图像.实验结果表明,该方法增强效果显著,不仅使图像边缘明显突出、纹理更加清晰,同时颜色保真,其视觉效果优于传统的图像增强方法,客观评价指标也有较大幅度的提升.

     

    Abstract: To make the detail of color images outstanding, highly visible, an image enhancement method based on the quaternion guided filter was proposed. First, the traditional guided filter was extended to quaternion field, and the guided filter based on quaternion was obtained. Then, the original color image was modeled in a quaternion matrix form, and the quaternion-value of the image was processed through quaternion guided filter. Furthermore, image layering was adopted to decompose a source image into a base image and a detail image, and the detail image was enhanced by self-adaptive enhancement transform. Finally, the base image and the detail-enhanced image were reconstructed to obtain an enhanced image. Experimental results show that the proposed method can produce high-quality detail enhancement. It not only makes image's edges more prominent and textures clearer, but also keeps color fidelity. Comparing with the existing enhancement methods, our proposed method improves the visual quality significantly and the objective evaluating indicators are also greatly improved.

     

/

返回文章
返回