共查询到19条相似文献，搜索用时 265 毫秒
1.
《林产化学与工业》2015,(6)
采用近红外光谱结合化学计量学的方法,对桉木和相思木及其属间6种木材的判别分类进行了研究。首先采集了尾巨桉、尾叶桉L11、尾叶桉U6、蓝桉、马占相思、厚荚相思,共计86个样本的近红外光谱图,采用偏最小二乘法判别分析(PLS-DA)建立了桉木和相思木的分类模型,校正集和验证集的预测值与实际值之间的回归线基本重合,决定系数(R2)分别为0.99和0.97,模型效果较好,且对未知样本的识别正确率为100%。为了对属间的6种木材作进一步的判别,采用MSC和Savitzky-Golay平滑对4000~7500 cm-1光谱进行预处理后,结合主成分分析(PCA)建立判别模型,模型识别率和验证正确率均为100%。结果表明基于近红外光谱结合化学计量学算法可以对桉木和相思木的不同属进行快速鉴别。 相似文献
2.
【目的】探究三维卷积神经网络（3D-CNN）在高光谱数据支持的树种分类中的有效网络构建方式，以提高树种分类精度。【方法】以美国加利福尼亚州内华达山脉南部为研究区，LiDAR数据获取的森林冠层高（CHM）进行单木分割并以此为补充建立样本，改进一种结构更简单、分类精度更高且无需对高光谱数据进行预处理的3D-CNN网络结构用于森林树种识别。【结果】相较于常规机器学习分类方法【支持向量机（SVM），随机森林（RF）】、传统二维卷积神经网络模型（2D-CNN）及最新多光谱分辨率三维卷积神经网络（MSR 3D-CNN）模型，本研究提出的3D-CNN模型对树种总体分类精度为99.79%，平均交并比（MIoU）为99.53%。与SVM和RF分类结果相比，本研究构建的3D-CNN模型总体分类精度提高5%左右，且具有对树种边界提取更加准确、椒盐现象更少发生的特点；与2D-CNN相比，总体分类精度提高10%左右，MIoU提高7%左右；与MSR 3D-CNN相比，总体精度相差不大，但在训练和测试过程中，本模型耗时远远小于MSR 3D-CNN模型。【结论】本研究改进的3D-CNN模型结构能够高效对原始高光谱影像... 相似文献
3.
以云南省香格里拉市为研究区,对ASD光谱仪实测的4种针叶树种光谱数据采用包络线去除法、光谱一阶微分法和光谱二阶微分法3种波段选择方法得到Hyperion高光谱影像数据的分类特征波段,采用最大似然法、支持向量机2种分类方法对所选的特征波段开展树种识别分类,对原始影像采用光谱角填图分类方法作对比实验。结果表明,基于ASD数据的光谱一阶波段选择方案的支持向量机分类方法精度最高,总体分类精度为81.95%,Kappa系数为0.725 1。采用ASD实测光谱数据能有效指导Hyperion进行树种分类,基于数据尺度和换算方式,一阶微分更适合特征波段选择;与传统的数理统计分类方法和光谱特征分类方法相比,基于机器学习的方法如支持向量机等在高光谱遥感分类中具有更大的应用潜力。 相似文献
4.
树种识别一直是困扰遥感研究的一个难点,而国产高分二号识别地物和树种具有巨大潜力。选取四川省甘孜州道孚县为研究区,利用高分二号4m多光谱遥感影像,并结合该县的森林资源二类调查结果数据,分别采用最大似然法和支持向量机方法,对利用高分二号数据在树种识别应用中的可能性进行探讨。研究结果表明:所采用的两种方法识别出研究区域主要树种的精度都高于80%,其中:采用最大似然法分类精度为81.79%,支持向量机方法分类精度为86.75%。在先验知识的支持下,利用高分二号多光谱影像也可用于树种识别研究中。 相似文献
5.
《林业科学》2021,57(8)
【目的】基于近红外光谱分析技术,判别纤维板生产过程中施胶量高低,为施胶工艺阶段提供技术支撑。【方法】以桉木纤维为研究对象,以脲醛树脂胶黏剂施胶量高低为判别指标,对同一批桉木纤维,控制含水率5%左右,建立施胶量为无(0%)、低(3%)、中(12%)、高(20%) 4种类别的近红外光谱偏最小二乘法回归模型(PLS),采用PLS-DA法判别施胶量高低,探究光谱采集方式(施胶纤维运动或静态状态)、施胶后纤维陈放时间对模型判别准确性的影响。【结果】1)随着施胶量增加,近红外光谱吸光度增大,PCA分析可区分无、低、中、高4种施胶量类别的桉木纤维; PLS-DA法能够建立相关性好、准确性高的近红外光谱模型,对未知样品的判别正确率达100%; 2)静态或动态光谱采集方式不会影响模型建立,对未知样品的判别正确率达100%;施胶后纤维陈放时间对近红外光谱影响很大,施胶后长时间陈放的纤维几乎不能正确判别施胶量高低。【结论】不同施胶量桉木纤维对近红外光谱的吸光度不同,PLS-DA法可准确判别桉木纤维施胶量高低;样品处于静止或运动状态均不影响模型建立和判别的准确性,可为在线识别纤维施胶量提供一定思路。 相似文献
6.
南方主要针叶树种高光谱数据降维分类研究 总被引：1，自引：0，他引：1
《中南林业科技大学学报(自然科学版)》2010,(11)
采用ASD公司生产的FieldSpec HandHeldTM地物光谱仪,分别于2005、2006、2008年冬季跟踪观测杉木、马尾松、黑松、雪松等针叶树种的高光谱数据,经筛选后获取有效观测数据160条,其中120条作为训练集,40条作为测试集。将平滑去噪的一阶微分高光谱数据进行PCA方法和GA方法降维,然后利用BP神经网络和支持向量机(SVM)对降维后的测试集数据进行分类。结果表明:PCA—BP神经网络模型分类准确率95%,PCA—SVM分类准确率97.5%,GA和BP分类准确率92.5%,GA-SVM分类准确率100%。这说明两种降维方式结合支持向量机的分类均优于其与BP神经网络结合的分类,基于GA的降维方法对高光谱波段的选择更有效率,具有较好的应用前景。 相似文献
7.
《中南林业调查规划》2017,(3)
以北京市西山试验林场为研究区域,利用Worldview—2影像构建各树种的光谱特征、地形特征、植被指数特征、纹理特征以及形态特征,建立关于山地森林树种识别的知识。采用基于像元和面向对象的方法进行树种识别分类。在基于像元的分类方法中,选择决策树分类和支持向量机分类;在面向对象的分类方法中,选择基于边缘检测的方法分割影像,用最近邻法分类。决策树分类的总体分类精度为65.62%,Kappa系数为0.588 9;支持向量机分类的总体分类精度为62.42%,Kappa系数为0.552 8;面向对象的分类方法总体分类精度为64.27%,Kappa系数为0.580 2。 相似文献
8.
《林业资源管理》2019,(5):44-51
树种分布是森林资源监测的一个重要指标,也是遥感影像在森林资源监测应用中的难点之一。基于国产高分二号卫星影像数据、森林资源二类调查数据、DEM数据,结合光谱、纹理、指数及地形因子等多种特征,比较支持向量机、随机森林和XGBoost等3种分类算法,根据分类精度选择最优算法(即XGBoost)进行特征筛选,对龙泉市的阔叶树、马尾松、杉木和毛竹等4种主要优势树种进行分类。结果表明:采用XGBoost分类模型的分类总精度为83.88%,Kappa系数0.78,较支持向量机和随机森林分类方法有明显提高。经特征选择后,虽未明显提高树种分类精度,但可以减少特征的冗余,为小样本数据下特征的选取降维提供了一定的参考。 相似文献
9.
10.
[目的]通过机载遥感影像对普洱山区进行植被分类研究,为山区森林经营规划与可持续经营方案的制图提供高效应用途径。[方法]将2014年4月航拍的机载AISA Eagle II高光谱和Li DAR同步数据融合,利用点云数据提取的数字冠层高度模型(CHM)得到树种的垂直结构信息,结合经过主成分分析(PCA)的高光谱降维影像,选用支持向量机(SVM)分类器进行分类。[结果]普洱市万掌山实验区主要树种分为思茅松、西南桦、刺栲、木荷等。融合影像数据分类的总体精度和Kappa系数分别为80.54%、0.78,比单一高光谱影像数据分类精度分别提高6.55%、0.08,其中主要经营树种思茅松的制图精度达到了90.24%。[结论]该方法对山区主要树种的识别是有效的,将机载Li DAR与高光谱影像融合可以有效改善分类精度。 相似文献
11.
为提高木质粉尘火花检测的准确性,利用基于支持向量机(SVM)的物质分类方法检测木质粉尘火花。选取马尾松和杨木粉尘为研究对象,将两种粉尘分组点燃试验,获取火花和灰分的高光谱图像,提取感兴趣区域(region of interest,ROI)内的发光度数据进行预处理。利用感兴趣区域内的数据建立SVM分类模型,分别利用网格搜索法(GS)、遗传算法(GA)以及粒子群算法(PSO)对两类树种的SVM分类模型进行参数优化,并将三种参数优选方法的分类预测准确率进行对比。结果表明,三种优化方法均能够很好地检测两种树种的木粉火花,其中网格搜索法检测准确率明显高于其余两种,更适于木质粉尘火花探测,这为人造板生产过程中能够高效检测木质粉尘火花提供了一定的理论依据。 相似文献
12.
Near Infrared (NIR) and Fluorescence (FS) spectroscopy were investigated for their ability to rapidly separate three Canadian softwoods: balsam fir, western hemlock, and white spruce. NIR and FS spectral data were used to develop classification models using soft independent modeling of class analogies (SIMCA) method. For each wood species, spectra of 90 wood specimens were collected over a wavelength window of 800–2,500?nm for NIR spectral data and a wavelength range of 380–540 and 380–705?nm for FS spectral data. Raw spectra and first-derivative-transformed spectra were used to develop NIR calibration models to separate the three wood species using the wavelength ranges, 800–2,500, 1,100–2,200, and 1,300–2,000?nm, by the SIMCA method. Similarly, FS raw spectral data were also used to develop FS calibrations using wavelength ranges of 380–540 and 380–705?nm. Principal component analysis models were made for each class from the calibration set consisting of 65 specimens of each of the three wood species. Specimens not present in the calibration set (27 specimens of each wood species) were tested for classification according to the SIMCA method at a 5 and 25% significance level. Type I error associated with the models developed with NIR spectral data ranged from 0 to 19 and 0 to 52% for the 5 and 25% significance levels, respectively, while type II error ranged from 2 to 50 and 0 to 19%, respectively. When tested at a 5% significance level, there was no significant improvement in NIR models developed with first-derivative-transformed spectra over models developed with raw spectra. Type I error associated with the models developed with Fluorescence spectral data ranged from 0 to 4 and 7 to 30% for the 5 and 25% significance levels, respectively, while type II error ranged from 1 to 9 and 0 to 1%, respectively. There were no significant differences in performance of FS models developed with spectra using wavelength ranges of 380–540 and 380–705?nm. 相似文献
13.
Quality assessment of thermally modified wood has evolved as one of the major fields in the research on thermal modification of wood. This study investigates NIR spectroscopy in combination with the pattern recognition method of soft independent modeling of class analogies (SIMCA). Focus is put on identifying different treatment intensities of thermally modified samples of beech, ash, and Norway spruce. The results indicate that SIMCA classification based on NIR spectroscopy could be used for quality control of thermally modified wood. The method might be applicable for producers (pre-delivery checks) and customers (reception control). However, transfer from laboratory to industrial conditions needs further investigation. 相似文献
14.
利用遥感数据开展森林资源优势树种的分类对森林资源的监测、森林可持续经营及生物多样性研究具有重要意义。研究针对复杂地形区域的破碎化森林,采用高分二号(GF-2)的多光谱影像作为基础数据进行森林优势树种的精细分类。本文以地形复杂、森林破碎化的湖北省竹山县九华山林场为研究对象,采用面向对象分类方法对树种进行精细分类,比较支持向量法、最近邻法(KNN)和随机森林(RF)三种不同分类算法的分类效果。在尺度阈值为30、合并阈值为95时分割的基础上,利用SVM、KNN和RF分类结果和分类精度差异较大。分类精度最高的是SVM分类方法,总体精度为68.52%,Kappa系数为0.62;其次为随机森林分类法,总体精度为60.29%,Kappa系数为0.54;KNN分类方法精度最低,总体精度为59.41%,Kappa系数为0.53。GF-2号数据能满足树种分类基本需求,在复杂地形和景观破碎化地区用支持向量机进行树种的分类精度更高,但仍存在一定的局限性。 相似文献
15.
Application of near-infrared spectroscopy to wood discrimination 总被引：4，自引：0，他引：4
This study deals with a new nondestructive discriminant analysis by which wood can be classified on the basis of a combination
of near-infrared (NIR) spectroscopy and Mahalanobis' generalized distance. Its accuracy and reasonability were examined for
wood samples with various moisture contents ranging from oven-dried to a fully saturated free water state. In a discriminant
analysis employing second derivative spectra, each wood group was well distinguished. Mahalanobis' generalized distances between
softwoods are relatively independent of analytical pattern, whereas the distances between hardwoods are large for easy classification.
There may be two reasons for selecting a wavelength: (1) when the chemical component of wood substance relates to the discriminant
analysis; and (2) when the difference in moisture content with wood species relates to them. When we correctly construct the
database of NIR spectra, confirming the purpose of the analysis, suitable wood discrimination should be possible.
Received: January 23, 2002 / Accepted: March 15, 2002
Acknowledgment The authors sincerely thank the Tanabe Southeast Asia Nations Friendship Foundation for financial support.
Part of this report was presented at the 51nd Annual Meeting of the Japan Wood Research Society, 2001
Correspondence to:S. Tsuchikawa 相似文献
16.
一种新的针叶材自动识别方法 总被引：1，自引：1，他引：0
提出通过横切面显微图像对针叶材树种进行计算机识别的方法。该方法通过提取图像的PCA特征,生成特征树,然后采用SVM对样本进行分类。使用8种针叶材,每种12个样本,并采用留一交叉验证,对图像的分割方法、最近邻与SVM分类算法和不同范数距离下的识别效果进行试验。结果表明通过部分木材微观的纹理结构进行木材识别的可能性。 相似文献
17.
An automated wood texture recognition system of 48 tropical wood species is presented. For each wood species, 100 macroscopic texture images are captured from different timber logs where 70 images are used for training while 30 images are used for testing. In this work, a fuzzy pre-classifier is used to complement a set of support vector machines (SVM) to manage the large wood database and classify the wood species efficiently. Given a test image, a set of texture pore features is extracted from the image and used as inputs to a fuzzy pre-classifier which assigns it to one of the four broad categories. Then, another set of texture features is extracted from the image and used with the SVM dedicated to the selected category to further classify the test image to a particular wood species. The advantage of dividing the database into four smaller databases is that when a new wood species is added into the system, only the SVM classifier of one of the four databases needs to be retrained instead of those of the entire database. This shortens the training time and emulates the experts’ reasoning when expanding the wood database. The results show that the proposed model is more robust as the size of wood database is increased. 相似文献
18.
This study deals with the suitable discriminant techniques of wood-based materials by means of near-infrared spectroscopy
(NIRS) and several chemometric analyses. The concept of Mahalanobis' generalized distance, K nearest neighbors (KNN), and
soft independent modeling of class analogy (SIMCA) were evaluated to determine the best analytical procedure. The difference
in the accuracy of classification with the spectrophotometer, the wavelength range as the explanatory variables, and the light-exposure
condition of the sample were examined in detail. It was difficult to apply Mahalanobis' generalized distances to the classification
of wood-based materials where NIR spectra varied widely within the sample category. The performance of KNN in the NIR region
(800–2500 nm), for which the device used in the laboratory was employed, exhibited a high rate of correct answers of validation
(>98%) independent of the light-exposure conditions of the sample. When employing the device used in the field, both KNN and
SIMCA revealed correct answers of validation (>88%) at wavelengths of 550–1010 nm. These results suggest the applicability
of NIRS to a reasonable classification of used wood at the factory and at job sites.
Received: March 13, 2002 / Accepted: July 19, 2002
Acknowledgments The authors thank Gifu Prefectural Human Life Technology Research Institute and Kubota Co. for their support. We also thank
Professor Dr. Shiro Kimura and Dr. Hideyuki Yokochi for their constructive discussions about the research. 相似文献
19.
To determine the independent decomposition rates of lignin and cellulose of decayed woody debris, a technique for the rapid analysis of lignin and cellulose is required. We applied a near-infrared spectroscopy (NIRS) technique to measure the lignin and holocellulose content in decayed wood. We succeeded in creating partial least-squares (PLS) models to estimate the lignin and holocellulose content in the decayed wood of five species using NIR spectra. Although the accuracy was acceptable for the estimation of a five-species mixed model (R 2 = 0.970 for lignin and R 2 = 0.962 for holocellulose), it was further improved when the model was applied to each species independently. This combination of NIRS and a PLS model is a valuable tool for the determination of the lignin and holocellulose content in decayed wood. The technique is time efficient (3 min per sample) and non-hazardous (no acid treatment is required). 相似文献