多传感器图像融合的客观评价与分析

图像融合作为一种有效的信息融合技术, 已经广泛用于军事、遥感、机器视觉和医疗诊断等领域。这里研究了多传感器图像 融合效果的客观评价方法, 其中包括均方根误差、平均误差、灰度标准差、熵、熵差、交叉熵、互信息和空间频率。将加权平均、金字 塔、小波变换三种方法用于多传感器图像的融合, 实验结果表明客观评价结果与视觉评价的结果是一致的, 是一种行之有效的评价 手段。
关键词　图像融合　小波变换　金字塔变换　客观评价
Abstract　Being an eff icien t m ethod of info rm at ion fu sion, im age fu sion has been u sed in m any f ields such as m ach ine vision, m edical diagno sis, m ilitary app licat ion s and remo te sen sing. The ob ject ive evaluat ion of fu sed im age is studied. The ob ject ive evaluat ion m easu res are roo t of m ean square erro r, m ean erro r, standard devia2 t ion, en t ropy, en t ropy difference, cro ss en t ropy, m u tual info rm at ion and spat ial f requency. Th ree im age fu2 sion algo rithm s, w eigh ted m ean, L ap lacian pyram id based, w avelet decompo sit ion based, are app lied to fu se several im ages. Experim en tal resu lt s show that conclu sion s of ob ject ive evaluat ion co incide w ith tho se of hu2 m an vision evaluat ion. So the ob ject ive evaluat ion m easu res are feasib le and eff icien t.
Key words　Im age fu sion　W avelet t ran sfo rm 　Pyram id t ran sfo rm 　O b ject ive evaluat ion