紫金山地区盗伐立木三维图像特征的远程可视化研究

Remote Visualization of Three-dimensional Image of Standing Purple with Illegal Logging Feature in Zijinshan Mountain Area

在远程可视化紫金山地区盗伐立木三维图像特征识别的研究中,传统的图像特征识别方法忽略了紫金山地区所处复杂环境的影响,从而导致对此地区的可视化研究局限于二维局部细节特征,无法实现远程可视化精准识别。针对此问题提出了一种基于特征转化的立木图像纹理特征识别方法,分析紫金山地区遭盗伐区域的背景环境,分析和现存的立木图像二维信息,并将采集得到的砍伐地区立木二维图像纹理特征矩阵实现迭代分解,在此基础上将不能够识别的立木边缘缺陷特征转化成可识别的三维图像纹理特征,最终实现远程可视化识别。实验结果表明,相比传统方法,本文提出的方法应用于在远程可视化系统识别方面具有更高的精准度。

Research on remote visualization Purple Mountain Area illegal felling standing threedimensional image feature recognition problem. The traditional image feature recognition approach ignores the complex environment in which the influence of the Purple Mountain region led visualization of this area has been limited to two-dimensional local minutiae, unable to achieve accurate remote visual recognition. Proposed stumpage image texture feature recognition method based on feature conversions,by analyzing the background was the Purple Mountain Area Regional Environment illegal logging,analyzing existing stumpage dimensional image information. Standing two- dimensional image texture features obtained felled areas will be collected to achieve iterative matrix decomposition, on this basis will not be able to identify characteristics of stumpage edge defects transforming into a threedimensional image texture features can be identified. Ultimately remote visualization recognition.Experimental results show that the proposed method is higher compared to the traditional method used in remote visualization system identification accuracy.