论文部分内容阅读
梯田作为黄土高原最为典型的人工地貌之一,有重要的农业生产和水土保持价值。传统的梯田自动提取仅限于梯田所在区域的范围划定,未能对梯田田坎线实现有效的自动化提取。鉴于此,本文提出了一种基于光照晕渲模拟的梯田快速提取方法。首先,对无人机航测生成的1 m分辨率的数字高程模型(DEM)进行4个方向的光照晕渲模拟,并相加取平均值;然后,通过适当的阈值对均值图像进行二值化,并掩膜掉沟谷等非梯田区域;最后,基于二值化图像自动矢量化得到梯田田坎线,并通过适当的长度阈值进行碎线过滤提高提取精度。本文以陕西省长武县王东沟流域为实验样区进行了实验,结果表明,该方法提取结果的准确率为89.09%,具有较好的提取精度。此外,对该方法涉及到的参数进行了讨论,表明光照模拟的方向角采用2个正交的对称方向对、高度角采用田坎坡度的反正切值、二值化阈值采用t=180-σ的经验公式,可以满足黄土高原的梯田自动快速提取。
As one of the most typical artificial landscapes in the Loess Plateau, terraced fields have important agricultural production and water and soil conservation values. Conventional terraced automatic extraction is limited to the area where terraced areas are located, failing to achieve effective automated extraction of terraced fields. In view of this, this paper presents a terraced rapid extraction method based on light shading rendering. Firstly, 4-directional light shading rendering is simulated on the 1-m digital elevation model (DEM) generated by the UAV’s aerial survey and averaged. Then, the mean value image is binarized by the appropriate threshold , And masked the non-terraced areas such as valleys. Finally, terraced ridge lines were obtained based on binary vectorization of automatic vectorization, and the broken lines were filtered by proper length threshold to improve the extraction accuracy. The experiment was carried out in Wangdonggou watershed of Changwu County, Shaanxi Province. The results showed that the accuracy of the method was 89.09%, and the extraction accuracy was good. In addition, the parameters involved in this method are discussed. It is shown that the direction of illumination simulation uses two orthogonal pairs of symmetric directions, the height angle adopts the arc tangent of ridge slope, and the binarization threshold adopts t = 180-σ The empirical formula can automatically and quickly extract terraces in the Loess Plateau.