一种快速提高林区影像可读性的算法

A Method to Improve the Readability of Forest Image

在摄影测量应用于森林资源调查的过程中,由于南方特有的低矮灌木遮挡,使得数码摄影成像的图像常常处于阴暗背景中,不利于目视判读,从而造成树干边缘提取的结果变得不准确.图像增强的传统方法,已经可以达到比较好的效果,但是林区图像背景复杂、色调单一.针对这一特征,从可读性方面进行了改进.通过改进PDE的相关算法,提出一种同时进行背景平滑和目标增强的方法,使得处理效率上比传统的方法提高80%～93.3%.使用ISCRl和ISCRg对结果进行辅助对比,发现该方法在噪声去除处理方面还有待提高.

In the application of photogrammetric surveying in investigating forest resources,due to the low bush of China＇s Southern Regions,the target are usually against dark backgrounds,thus the difficulty in visual interpretation,which makes the obtained result of the tree outline inaccurate.Traditional methods of image enhancement have achieved relatively good results,but the forest image has its characteristics,which has complex background and single color.In this paper,efforts was made to improve the readability of the forest images.By improving the PDE model,a method which can smooth background and enhance objectives at the same time is raised,which will increase the effect than the traditional method by 80~93.3%.ISCRl and ISCRg were used to help explain the results.However,the noise removal processing still needs improving.