便携式测树仪的研制及应用研究

发布时间：2018-05-08 21:49

本文选题：测树仪 + CMOS　；参考：《北京林业大学》2016年博士论文

【摘要】：在森林资源调查中需要调查林分因子,从而了解森林资源的数量与质量,以进行林业规划、设计和经营。森林资源调查是一项费时费力的工作,使用传统测量工具需要不同仪器来进行测量,同时还受地形条件的限制。我国常用直径测量方式为围尺,但其无法测量上部直径,并且在复杂环境下人难以行进的地方也难以测量。测高仪器主要为布鲁莱斯测高器,需测量水平距,受地形条件限制大,误差较大。为快速精确进行林木测量,开发设计了便携式数字化智能测树仪—Mini测树仪。仪器由STM32F103VET6单片机、ov9655传感器、LIS331DLTR倾角传感器、键盘、LTTD240320022-L3-TF显示屏模块及其相关电路模块等构成。采用相位原理测距,测距的平均误差为0.047m。在径高同测模式下,利用摄像头和CMOS传感器对图像进行处理,实现树径的自动识别,再利用所测距离和倾角,实现高度、直径同步测量。另外仪器还有测径模式、测高模式、测距模式、倾角测量模式、面积计算模式。将C言语编写的内置代码烧入单片机中,实现了各种测量,再配合后处理软件,实现蓄积计算、林木分布图出图以及竞争指数的计算等功能,使得内外业一体化。通过在北京的奥林匹克森林公园、鹫峰以及福建省连城县朋口镇进行了应用和精度验证,本研究主要内容和成果如下：(1)实现树木直径的自动判别和测量直径自动测量方面通过摄像头和CMOS传感器获得图像,经灰度转换后,使用窗口大小为5的高斯逆波器进行平滑、非极大值剔除、双阀值的边缘连接等一系列Canny算法,提取边缘。在提取的边缘图像基础上,设计了树木直径自动提取算法,再通过焦距、物距、像距和像素大小之间的关系计算出树干直径。选择了不同树木进行验证,实验结果表明树干识别的精度为87.9%,树径测量的自动化程度高,所测树干直径值一半以上误差在5%以内,大部分测量数据符合森林资源调查需求。在自动识别与测量的基础上加入手动调节功能,对识别效果不好的进行调节从而提高树木直径测量的精度。本仪器解决了给定高度处的直径测量,即实现树高、直径同步测量,特别是胸径地测量。径高同步测量主要通过激光测距传感器、倾角传感器和树径测量功能来实现。对上部直径进行测量验证,对于高20m以上的树,Mini测树仪的误差从高度上说,随着树高的增加而增加,1-5m处最小误差为0.37cm,5-10m为0.56cm,10-15m为0.58cm,15-20m为0.58cm,20m以上为0.83cm。而从距离上看,误差小的一般在最后两个距离段区间内。13m以下树木测量中,Mini测树仪1-2m处最小误差为0.15cm,2-3m处最小误差为0.22cm,3-5m为0.46cm,5-8m为0.54cm,8-13m为0.57cm。对于上部直径被遮挡的情况,使用削度方程进行估算,只需测量胸径,然后就可获得所需高度处的直径。使用曾伟生(1997)和Mei(2015)的削度方程进行了拟合,得到曾伟生的精度高,相对误差在9%内,而Mei的误差较大。(2)实现树高的快速方便测量。树高测量方面,基于三角函数的测高主要使用测距仪、倾角传感器实现。CMOS传感器测高主要由RGB至HIS的转换、树干细化、投影变换、投影重建等步骤实现。在高度测量方面,对于胸高处的测量,最大误差为0.012m,平均误差为1.298×10-3m。对不同倾角进行树高测量,得到：35°-60°之间所测树高相对误差几乎都在5%以内。大于60°和小于35°的误差较大,60°以上测量的平均相对误差为10.43%,35°-60°的相对误差为2.53%,0°-35°的相对误差为5.65%,可知在35°-60°间测量树高能准确的测出树高。(3)实现林地和林分的各种测量由Mini测树仪的各种功能实现冠幅测量、立木体积测量、郁闭度测量、数字角规测量、圆形样地测量、中心树N边行样地法测量、竞争指标测量等。其中数字角规测量实现了林分蓄积、株数密度测量等。在林地和林分因子测量方面,对于冠幅的测量,将Mini测树仪和皮尺测量值作对比,两者的绝对误差最大的为0.494m,RMSE为0.283m,精度为96.6%。立木体积测量,一共测量了82棵树的立木体积,Mini测树仪与卡尺所测最大误差为13.34%。数字角规测量方面,人工林的精度高,其中第五块样地4.8系数精度最大,比实际多了25棵,相对误差为1.7%。其次为第一样地的系数1,相对误差为3.7%,其他误差都在10%以上。而天然林更复杂,内部情况不一,误差在20%-40%间,误差高。数字角规的蓄积测量绝对误差最大值为19.02m3,最小误差为1.19m3。
[Abstract]:Forest Resource Survey needs to be investigated in order to understand the quantity and quality of forest resources so as to carry out forestry planning, design and management. Forest resource investigation is a time-consuming and laborious work. The use of traditional measuring tools is required by different instruments to measure and be restricted by the terrain conditions. But it can't measure the upper diameter, and it is difficult to measure the place where people can't walk in the complex environment. The height measuring instrument is mainly the Blus altimetry. It needs to measure the horizontal distance, the terrain conditions are limited and the error is large. The portable digital intelligent tree measuring instrument, Mini, is developed and designed for fast and accurate forest tree measurement. The instrument is composed of STM32F103VET6 single chip, OV9655 sensor, LIS331DLTR tilt sensor, keyboard, LTTD240320022-L3-TF display module and its related circuit modules. The average error of distance measurement is 0.047m. in the same measurement mode of height and height, and the image is processed by using the camera and CMOS sensor. The height and diameter of the tree are automatically identified, and the height and diameter are synchronized. In addition, the diameter measurement mode, the height measurement mode, the distance measurement mode, the angle measurement mode and the area calculation mode are used. The built-in code compiled by the C language is burned into the single chip microcomputer, and the measurement is realized, and the post processing software is combined with the storage calculation, and forest calculation is realized. The functions of the wood distribution map and the calculation of the competition index make the internal and external integration. The main contents and results of this study are as follows: (1) the automatic measurement of the diameter of trees and the measurement of the diameter of the tree are realized by the application and precision verification of the jiukou town in Beijing's Olympic, Jiufeng and Liancheng County of Fujian province. The image is obtained through the camera and CMOS sensor. After the gray scale conversion, a series of Canny algorithms, such as the Gauss inverse wave device with the size of 5, are used for smoothing, non maximum elimination and double threshold edge connection, and a series of Canny algorithms are used to extract the edges. The relationship between distance and pixel size is calculated. Different trees are selected to verify the tree trunk. The experimental results show that the accuracy of tree trunk recognition is 87.9%, the automation of tree diameter measurement is high and the diameter of tree trunk is less than 5%. Most of the measured data conform to the demand of forest resources investigation. On the basis of the manual adjustment function, the accuracy of the measurement of tree diameter is improved by adjusting the poor recognition effect. The instrument solves the measurement of diameter at a given height, that is, the height of the tree, the measurement of diameter synchronously, especially the diameter of the chest. The diameter height synchronization measurement is mainly through the laser range sensor, the tilt sensor and the tree diameter. Measure the function to realize. To measure the upper diameter, for the tree above 20m, the error of the Mini tree measuring instrument is from height, with the increase of tree height, the minimum error of 1-5m is 0.37cm, 5-10m is 0.56cm, 10-15m is 0.58cm, 15-20m is 0.58cm, 20m above is 0.83cm. and the small error is generally in the last two In the measurement of trees below.13m in the range section, the minimum error of the Mini tree measuring instrument at 1-2m is 0.15cm, the minimum error at 2-3m is 0.22cm, 3-5M is 0.46cm, and 5-8m is 0.54cm. The 8-13m is the case of the upper diameter of the 0.57cm., using the sharpening equation to estimate only the diameter of the chest, then the diameter of the required height is obtained. Used once. The sharpening equation of Weisheng (1997) and Mei (2015) is fitted, and the accuracy of Ceng Weisheng is high, the relative error is within 9%, and the error of Mei is larger. (2) the rapid and convenient measurement of tree height. The height measurement of the tree is mainly based on the trigonometric function, and the tilt sensor realizes the conversion of the.CMOS sensor mainly from RGB to HIS. In the height measurement, the maximum error is 0.012m, the average error is 1.298 x 10-3m. for the tree height measurement of different dip angles, and the relative error between 35 degrees -60 degrees is almost 5%. The error is greater than 60 and less than 35 degrees, and the error is larger, 60 degrees. The average relative error of the above measurement is 10.43%, the relative error of 35 degree -60 degree is 2.53% and the relative error of 0 degree -35 degree is 5.65%. It is known that the high energy and accurate tree height of the tree is measured between 35 degrees -60 degrees. (3) the various functions of the woodland and forest are measured by various functions of the Mini tree measuring instrument, the volume measurement of the tree, the measurement of the canopy density and the digital angle. Measurement, circular sample land measurement, Center tree N edge sampling method measurement, competition index measurement, etc. in which the digital angle gauge realizes the stand accumulation, plant number density measurement and so on. In the forest and stand factor measurement, the measurement of the crown width, the Mini tree meter and the measuring value of the ruler are compared, the absolute error of the two is 0.494m, RMSE For 0.283m, a total of 82 trees was measured by 96.6%. volume measurement. The maximum error of 82 trees was measured. The maximum error measured by the Mini tree meter and caliper was the 13.34%. digital angle gauge. The accuracy of the plantation was high, of which the 4.8 coefficients of the fifth plots were the largest, and 25 were more than the actual, and the relative error was the second factor of the second coefficient 1. The error is 3.7% and the other errors are above 10%. While the natural forest is more complex, the internal situation is different, the error is in the 20%-40%, the error is high. The maximum error of the digital angle gauge is 19.02m3, and the minimum error is 1.19m3.

【学位授予单位】：北京林业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S758.7

【相似文献】