利用量化判据进行多聚焦融合图像的重构分析

Reconstruction analysis of multifocus fused images using quantitative criteria

提出一种基于对比度视觉模型的图像融合自适应分块搜索算法,利用几种量化判据,即均方根误差、熵、交叉熵和互信息进行融合图像序列分析及效果评价,对同一场景两幅严格配准的多聚焦图像的清晰恢复进行了深入研究。通过大量的图像实验表明:无论是依据视觉的主观评价,还是客观评价准则,该方法能实现多聚焦图像对标准参考图像的精确重构或优化融合效果。

A reconstruction algorithm was developed for fusion of two spatially registered images of the same scene. Quantitative criteria such as root mean square error, entropy, cross entropy and mutual information were used on the analysis and evaluation of fused image series. The clear restoration of multifocus images based on contrast vision model was studied deeply exploiting adaptive searching blocks algorithm. Extensive experiments demonstrate that most image cases achieve accurate reconstruction or optimized fused result to the reference image whether subjective visual effect or objective evaluation criteria.