空间频率图像的智能融合

Intelligent Fusion of Image in Spatial Frequency

提出了一种智能地融合同一场景的多幅图像为一幅图像的方法,与原图像相比,产生的图像包含较少的噪声和更多的信息。首先对原始图像除去斑点噪声;然后,运用直方图均衡化表示图像的细节和最大化图像信息内容;第三步,运用金字塔将图像分解为子图像,利用图像的密度、空间频率等特征,寻找需要融合的图像的部分;第四步,对这些子图像进行配准,为融合做准备;第五步,对需要融合的子图像,计算每一幅的空间频率,对空间频率不同的配准子图像,将结合它们周围子图像的空间频率信息,给出这个子图像的频率(一般情况下,频率值选择一个具有最高值的像素);第六步,为了得到更多细节,用模糊插值方法扩大这幅子图像。

A process is designed to fuse multiple images in the same scene,and to produce an image that contains less noise and more inforrnstion. First, the speckle noise is removed. Second, histogram equalization is applied to expose details and maximize the information content of the image. Third, using pyramid representation to express the/mage, find the differences of the subimages by intensity and spatial frequency. Fourth, images are registered to prepare for fusion, in the fifth step,apply new pointwise spatial frequency methodology by computing it at each pixel in each image. The images are then fused at each output pixel location by comparing the pointwise spatial frequency values at that location in all images and selecting the pixel with the highest such value. The sixth step enlarges this image with fuzzy interpolation for more detail.