| 摘要: |
| 目的 为了解决当前图像融合算法在融合过程中忽略了低频系数中所包含的图像细节信息，导致其输出的融合图像存在间断以及模糊效应的不足，方法 提出基于二代Curvelet变换耦合二维因子的图像融合算法。首先，利用具有多尺度以及多方向特性的二代Curvelet变换对源图像进行快速的分解，以获取源图像精细的低频以及高频系数。引入低频系数的信息熵以及区域方差特征来构造二维因子，对低频系数所包含的信息量以及图像的变化程度进行度量，以完成低频系数的融合。随后，利用高频系数的平均梯度特征，构造信息融合规则，完成高频系数的融合，提高融合图像的细节信息含有量。最后，利用像素点的R，G，B值，构造颜色校正因子，对融合图像进行颜色修正，以获取色彩效果较好的融合图像。结果 实验结果显示，与当前图像融合算法相比，所提算法具有更强的细节表达能力，其输出的融合图像具有更好的清晰度及视觉效果。结论 所提算法拥有较好的融合质量，能提高图像的对比度与分辨率，在图像处理领域具有一定的参考价值。 |
| 关键词: 图像融合 二代Curvelet变换 信息熵 区域方差 平均梯度 二维因子 低频系数 |
| DOI：10.19554/j.cnki.1001-3563.2019.13.035 |
| 分类号:TP391 |
| 基金项目:吉林省青年科学基金（20160520011JH，20180520021JH） |
|
| Image Fusion Algorithm Based on Second-generation Curvelet Transform Coupled with Two-dimensional Factor |
|
HAN Ming1, LI Hong-tu2
|
|
1.Department of Computer Science, The Branch of Jilin Normal University, Siping 136000, China;2.College of Computer Science and Technology, Jilin University, Changchun 130012, China
|
| Abstract: |
| The work aims to solve the problems as discontinuity and blurring effect of the fused image induced by neglecting the image details contained in the low frequency coefficients in the current image fusion algorithm in the process of fusion. An image fusion algorithm based on two-dimensional factor coupled with second-generation curvelet transform was proposed. Firstly, second-generation curvelet transform with multi-scale and multi-direction characteristics was used to decompose the source image rapidly to obtain the fine low-frequency and high-frequency coefficients of the source image. Two-dimensional factors were constructed by introducing the information entropy of low-frequency coefficients and regional variance characteristics. The main information and change degree of the image contained in low-frequency coefficients were measured, in order to complete the fusion of low-frequency coefficients. Subsequently, by means of the average gradient characteristics of high-frequency coefficients, the information fusion rules were established to complete the fusion of high-frequency coefficients, in order to further improve the detail information content of the fused image. Finally, the R, G and B values of the pixels were used to construct a color correction factor to correct the fused image in order to obtain the fused image with better color effect. The experimental results showed that, the proposed method had better detail expression ability than the current image fusion method, and the fused image had better clarity and visual effect. The proposed algorithm has better fusion quality and it can improve the contrast and resolution of the image, which has a certain reference value in the field of image processing. |
| Key words: image fusion second-generation curvelet transform information entropy regional variance average gradient two-dimensional factor low-frequency coefficient |
