三江源典型区鼠洞无人机遥感识别研究
|
摘要
草原鼠洞的识别定位可以为鼠害的监测、预测和防治等提供科学参考,因此,如何快速且准确地识别草地鼠洞成为亟需解决的问题。选取玛多县典型区作为研究区,利用可见光波段无人机影像,研究并建立了面向对象的模板匹配法和支持向量机法的草地鼠洞自动识别方法。模板匹配法是在多尺度分割的影像中选取不同种类的鼠洞对象并生成匹配模板,接着进行目标检测并产生初始结果,最后构建光谱、几何和纹理特征库对检测结果进行筛选。支持向量机法首先采集鼠洞训练样本并优化分类特征空间,然后采用支持向量机分类器监督分类得到鼠洞识别结果。对2种方法的识别结果进行精度评价与分析表明:2种方法的总体精度均较高,适用于三江源区草原鼠洞的精准识别。基于面向对象的模板匹配法比支持向量机法的总体识别精度整体高1%,错分误差低3%,识别效果较好。
Identification and location of rat holes in grassland provide scientific reference for the monitoring,prediction and prevention of rodent. How to identify rat holes in grassland quickly and accurately becomes an urgent problem. Based on visible image acquired by unmanned aerial vehicles( UAVs) in a typical area of Madoi County,this study used the Object-oriented Template Matching( OTM)method and Support Vector Machine( SVM) method to identify its rat holes in grassland automatically.OTM method is to select different kinds of rat-hole objects from multi-scale segmentation images. Then it generates matching templates,target detection and initial results generated. Finally the spectral,geometric and texture features of the target rat-hole objects are extracted to construct a feature knowledge base and to screen the detection results. SVM method firstly collects the training samples and optimizes the classification feature space. Then SVM classifier is used to supervise the classification to get the rat-hole recognition result. The accuracy evaluation and analysis of the two methods show that they have high overall accuracy,and are suitable for the accurate identification of rat holes in grassland of"Three-River Headwaters"region. The overall identification accuracy of object-oriented template matching method is 1% higher than the other one,and the commission error is 3% lower showing slightly better effect of identification.
Identification and location of rat holes in grassland provide scientific reference for the monitoring,prediction and prevention of rodent. How to identify rat holes in grassland quickly and accurately becomes an urgent problem. Based on visible image acquired by unmanned aerial vehicles( UAVs) in a typical area of Madoi County,this study used the Object-oriented Template Matching( OTM)method and Support Vector Machine( SVM) method to identify its rat holes in grassland automatically.OTM method is to select different kinds of rat-hole objects from multi-scale segmentation images. Then it generates matching templates,target detection and initial results generated. Finally the spectral,geometric and texture features of the target rat-hole objects are extracted to construct a feature knowledge base and to screen the detection results. SVM method firstly collects the training samples and optimizes the classification feature space. Then SVM classifier is used to supervise the classification to get the rat-hole recognition result. The accuracy evaluation and analysis of the two methods show that they have high overall accuracy,and are suitable for the accurate identification of rat holes in grassland of"Three-River Headwaters"region. The overall identification accuracy of object-oriented template matching method is 1% higher than the other one,and the commission error is 3% lower showing slightly better effect of identification.
引文
[1]王加亭,贠旭江,苏红田,等.“三江源”地区退化草原的鼠害监测技术[J].草业科学,2008,25(8):110-112.
[2]程军. 2002年我国草原鼠害发生状况及今后的防治策略[J].西北农业学报,2003,12(3):115-118.
[3]WANG W,FENG Q S,HUI Y U,et al. Application of“3S”technology on monitoring and prediction of the grassland rodents and insects[J]. Pratacultural Science,2010,27(3):31-39.
[4]杨东生,张生祥,阿帕尔,等.鼠害的防治[J].新疆畜牧业,2008,(3):58-60.
[5]鲁鸿佩,孙爱华,马绍慧.临夏州太子山麓天然草地鼠害调查及防治[J].草业科学,1999,16(5):30-34.
[6]李成文.青海地区草原鼠害的防治——以“三江源地区”为例[J].中国畜牧兽医文摘,2017,33(9):19-19.
[7]杨彦东,苏军虎,花立民.碌曲县草原鼠害区划研究[J].草原与草坪,2014,34(6):51-55.
[8]何咏琪,黄晓东,侯秀敏,等.基于3S技术的草原鼠害监测方法研究[J].草业学报,2013,22(3):33-40.
[9]徐正刚,赵运林,李波,等.基于MODIS植被指数评估洞庭湖区东方田鼠大暴发的危害[J].生态学报,2014,34(23):7101-7109.
[10]黄建文,鞠洪波,特木钦,等.阿拉善左旗天然梭梭林鼠害防治的遥感监测[J].林业科学,2004,40(3):107-110.
[11]VOGT J T,WALLET B. Feasibility of using template-based and object-based automated detection methods for quantifying black and hybrid imported fire ant(Solenopsis invicta and S. invicta×richteri)mounds in aerial digital imagery[J]. Rangeland Journal,2008,30(3):291-295.
[12]SANDINO J,WOOLER A,GONZALEZ F. Towards the automatic detection of pre-existing termite mounds through UAS and hyperspectral imagery[J]. Sensors,2017,17(10):2196.
[13]尼玛卓玛.三江源鼠害成因分析及治理[J].中国畜牧兽医文摘,2015,(2):32-32.
[14] DRGUT L,CSILLIK O,EISANK C,et al. Automated parameterisation for multi-scale image segmentation on multiple layers[J]. Isprs Journal of Photogrammetry&Remote Sensing,2014,88(100):119-127.
[15]罗巍,邵全琴,王东亮,等.基于面向对象分类的大型野生食草动物识别方法——以青海三江源地区为例[J].野生动物学报,2017,38(4):561-564.
[16]周在明,杨燕明,陈本清.基于无人机影像的滩涂入侵种互花米草植被信息提取与覆盖度研究[J].遥感技术与应用,2017,32(4):714-720.
[17]汪小钦,王苗苗,王绍强,等.基于可见光波段无人机遥感的植被信息提取[J].农业工程学报,2015,31(5):152-159.
[18]丁世飞,齐丙娟,谭红艳.支持向量机理论与算法研究综述[J].电子科技大学学报,2011,40(1):2-10.
[19]FEINSTEIN A R,CICCHETTI D V. High agreement but low kappa:I. The problems of two paradoxes[J]. Journal of Clinical Epidemiology,1990,43(6):543-549.
[20]CICCHETTI D V,FEINSTEIN A R. High agreement but low kappa:II. Resolving the paradoxes[J]. Journal of Clinical Epidemiology,1990,43(6):551-558.
[2]程军. 2002年我国草原鼠害发生状况及今后的防治策略[J].西北农业学报,2003,12(3):115-118.
[3]WANG W,FENG Q S,HUI Y U,et al. Application of“3S”technology on monitoring and prediction of the grassland rodents and insects[J]. Pratacultural Science,2010,27(3):31-39.
[4]杨东生,张生祥,阿帕尔,等.鼠害的防治[J].新疆畜牧业,2008,(3):58-60.
[5]鲁鸿佩,孙爱华,马绍慧.临夏州太子山麓天然草地鼠害调查及防治[J].草业科学,1999,16(5):30-34.
[6]李成文.青海地区草原鼠害的防治——以“三江源地区”为例[J].中国畜牧兽医文摘,2017,33(9):19-19.
[7]杨彦东,苏军虎,花立民.碌曲县草原鼠害区划研究[J].草原与草坪,2014,34(6):51-55.
[8]何咏琪,黄晓东,侯秀敏,等.基于3S技术的草原鼠害监测方法研究[J].草业学报,2013,22(3):33-40.
[9]徐正刚,赵运林,李波,等.基于MODIS植被指数评估洞庭湖区东方田鼠大暴发的危害[J].生态学报,2014,34(23):7101-7109.
[10]黄建文,鞠洪波,特木钦,等.阿拉善左旗天然梭梭林鼠害防治的遥感监测[J].林业科学,2004,40(3):107-110.
[11]VOGT J T,WALLET B. Feasibility of using template-based and object-based automated detection methods for quantifying black and hybrid imported fire ant(Solenopsis invicta and S. invicta×richteri)mounds in aerial digital imagery[J]. Rangeland Journal,2008,30(3):291-295.
[12]SANDINO J,WOOLER A,GONZALEZ F. Towards the automatic detection of pre-existing termite mounds through UAS and hyperspectral imagery[J]. Sensors,2017,17(10):2196.
[13]尼玛卓玛.三江源鼠害成因分析及治理[J].中国畜牧兽医文摘,2015,(2):32-32.
[14] DRGUT L,CSILLIK O,EISANK C,et al. Automated parameterisation for multi-scale image segmentation on multiple layers[J]. Isprs Journal of Photogrammetry&Remote Sensing,2014,88(100):119-127.
[15]罗巍,邵全琴,王东亮,等.基于面向对象分类的大型野生食草动物识别方法——以青海三江源地区为例[J].野生动物学报,2017,38(4):561-564.
[16]周在明,杨燕明,陈本清.基于无人机影像的滩涂入侵种互花米草植被信息提取与覆盖度研究[J].遥感技术与应用,2017,32(4):714-720.
[17]汪小钦,王苗苗,王绍强,等.基于可见光波段无人机遥感的植被信息提取[J].农业工程学报,2015,31(5):152-159.
[18]丁世飞,齐丙娟,谭红艳.支持向量机理论与算法研究综述[J].电子科技大学学报,2011,40(1):2-10.
[19]FEINSTEIN A R,CICCHETTI D V. High agreement but low kappa:I. The problems of two paradoxes[J]. Journal of Clinical Epidemiology,1990,43(6):543-549.
[20]CICCHETTI D V,FEINSTEIN A R. High agreement but low kappa:II. Resolving the paradoxes[J]. Journal of Clinical Epidemiology,1990,43(6):551-558.