面向产出的“高光谱遥感”线上线下混合式课程教学改革

“高光谱遥感”以其获取地物详细光谱数据的能力,为遥感信息定量应用开辟了新前景,是遥感专业高年级本科生巩固遥感原理与方法、开拓光学定量遥感应用新视野的重要课程。本文在“新工科”建设需求下,总结高光谱遥感课程教学经验,探索线上线下混合式课程改革方案,思考课程建设新举措。根据教学反馈结果,本文提出的教学改革思路不仅能激发学生的课程学习兴趣,还能提高学生的专业知识综合应用能力,为遥感科学与技术专业学生的深造学习和工程应用打下良好基础。

Yield Estimation Model of Citrus Based on Spectral Data and Agronomic Parameters

With the development of precision agriculture, the research that applies Remote Sensing technology, especially hyperspectral remote sensing, to realize crop management, monitoring and yield estimation, has been concerned. Nowadays, the growth-monitoring and yield-estimating methods in rice, wheat and other annual crops develop rapidly with some achievements having already been put into service. But the yield estimation research on perennial economic crops is few. Taking peren- nial citrus trees as the research object, using ASD spectrometer to collect citrus canopy spectral, this article studied and analyzed the citrus of veget＆tion index and its relationship on yield, synthetically considered the influence of the agriculture pa- rameters on crop yield, and finally constructed the citrus yield estimation model based on the spectral data and agronomic parameters. Through the Significance Test and Samples＇ Test, olutained that the model＇s fitting degree was R=0.631, F= 13.201, P〈0.01 and the error rate of estimating accuracy was controlled in the range 3%-16%, proving that the model has statistical signification and reliability. It concluded that hyperspectral acquired from citrus canopy has substantial potential for citrus yield estimation. This study is an application and exploration of Hyperspectral Remote Sensing technology in the citrus yield estimation.

Bathymetry and bottom albedo retrieval using Hyperion:a case study of Thitu Island and reef

The Spratly(Nansha) Islands in the South China Sea have considerable economic and important militarily strategic status.Ocean color remote sensing is an effective mean of surveying and research and especially it is useful for areas that are difficult to access,such as Thitu Island and its reef in the Spratly Islands.The Hyper-spectral Optimization Process Exemplar(HOPE) model,developed by Lee et al.(1999) is a rapid and robust bathymetry method that uses hyper-spectral remote sensing.In this study,using Hyperion hyper-spectral sensor data and HOPE,we derive bathymetry and bottom albedo measurements around Thitu Island and its reef.We compare the distribution of bottom depths from C-MAP with that derived from the Hyperion data.The retrieved bathymetry results correlate well with the distribution obtained from the bathymetry contour from 2.0 to 20 m.The average difference between Hyperion and C-MAP for two selected transects was 17.1%(n=59,R=0.848,RMSE=2.342) and 10.9%(n=59,R2=0.834,RMSE=0.463).The retrieved bottom albedo is homogeneous in the lagoon and significantly non-homogeneous around the lagoon.These results indicate that HOPE could be very useful for bathymetry studies for the islands of the South China Sea.

Minimum distance constrained nonnegative matrix factorization for hyperspectral data unmixing

This paper considers a problem of unsupervised spectral unmixing of hyperspectral data. Based on the Linear Mixing Model （ LMM）, a new method under the framework of nonnegative matrix fac- torization （NMF） is proposed, namely minimum distance constrained nonnegative matrix factoriza- tion （MDC-NMF）. In this paper, firstly, a new regularization term, called endmember distance （ED） is considered, which is defined as the sum of the squared Euclidean distances from each end- member to their geometric center. Compared with the simplex volume, ED has better optimization properties and is conceptually intuitive. Secondly, a projected gradient （PG） scheme is adopted, and by the virtue of ED, in this scheme the optimal step size along the feasible descent direction can be calculated easily at each iteration. Thirdly, a finite step （ no more than the number of endmem- bers） terminated algorithm is used to project a point on the canonical simplex, by which the abun- dance nonnegative constraint and abundance sum-to-one constraint can be accurately satisfied in a light amount of computation. The experimental results, based on a set of synthetic data and real da- ta, demonstrate that, in the same running time, MDC-NMF outperforms several other similar meth- ods proposed recently.

Classification of hyperspectral remote sensing images based on simulated annealing genetic algorithm and multiple instance learning

A hybrid feature selection and classification strategy was proposed based on the simulated annealing genetic algonthrn and multiple instance learning （MIL）. The band selection method was proposed from subspace decomposition, which combines the simulated annealing algorithm with the genetic algorithm in choosing different cross-over and mutation probabilities, as well as mutation individuals. Then MIL was combined with image segmentation, clustering and support vector machine algorithms to classify hyperspectral image. The experimental results show that this proposed method can get high classification accuracy of 93.13% at small training samples and the weaknesses of the conventional methods are overcome.

Spectral separability of tropical forest tree species using airborne hyperspectral imager

In Malaysia, airborne hyperspectral remote sensing is a relatively new technique used for research and commercial value in forest inventory and mapping. An advantage of airborne remote sensing, compared to satellite remote sensing, is its capability of offering a very high spatial resolution images. Thus, UPM-TropAIR AISA＇s airborne hyperspectral imagery that has been used in this study provides great quantity, better quality and also lower cost in identifying, quantifying and mapping of the Malaysian tropical timber forest resources. For the first stage in this study, the development of spectral library is deemed required in order for the Spectral Angle Mapper （SAM） classification be used to separate and map individual tree species in a tropical mixed mountain forest of Gunong Stong Forest Reserve. Pre-processing, enhancement and interpretation of image were conducted using ENVI Version 4.0 software. Results indicated that a total of eight commercial timber tree species was identified and mapped in a study plot of 5 ha using the TropAIR airborne hyperspectral imager with the aid of ground truthings.

Model-Based Integrated Methods for Quantitative Estimation of Soil Salinity from Hyperspectral Remote Sensing Data:A Case Study of Selected South African Soils

Soil salinization is a land degradation process that leads to reduced agricultural yields. This study investigated the method that can best predict electrical conductivity (EC) in dry soils using individual bands, a normalized difference salinity index (NDSI), partial least squares regression (PLSR), and bagging PLSR. Soil spectral reflectance of dried, ground, and sieved soil samples containing varying amounts of EC was measured using an ASD FieldSpec spectrometer in a darkroom. Predictive models were computed using a training dataset. An independent validation dataset was used to validate the models. The results showed that good predictions could be made based on bagging PLSR using first derivative reflectance (validation R2 = 0.85), PLSR using untransformed reflectance (validation R2 = 0.70), NDSI (validation R2 = 0.65), and the untransformed individual band at 2257 nm (validation R2 = 0.60) predictive models. These suggested the potential of mapping soil salinity using airborne and/or satellite hyperspectral data during dry seasons.

Estimating Leaf Chlorophyll Content Using Red Edge Parameters

Hyperspectral remote sensing makes it possible to non-destructively monitor leaf chlorophyll content （LCC）. This study characterized the geometric patterns of the first derivative reflectance spectra in the red edge region of rapeseed （Brassica napus L.） and wheat （Triticum aestivum L.） crops. The ratio of the red edge area less than 718 nm to the entire red edge area was negatively correlated with LCC. This finding allowed the construction of a new red edge param- eter, defined as red edge symmetry （RES）. Compared to the commonly used red edge parameters （red edge position, red edge amplitude, and red edge area）, RES was a better predictor of LCC. Furthermore, RES was easily calculated using the reflectance of red edge boundary wavebands at 675 and 755 nm （R675 and R755） and reflectance of red edge center wavelength at 718 nm （R718）, with the equation RES = （R71s - R675）/（R755 - R675）. In addition, RES was simulated effectively with wide wavebands from the airborne hyperspectral sensor AVIRIS and satellite hyperspectral sensor Hyperion. The close relationships between the simulated RES and LCC indicated a high feasibility of estimating LCC with simulated RES from AVIRIS and Hyperion data. This made RES readily applicable to common airborne and satellite hyperspectral data derived from AVIRIS and Hyperion sources, as well as ground-based spectral reflectance data.

Classification of hyperspectral remote sensing images using frequency spectrum similarity

An algorithm of hyperspectral remote sensing images classification is proposed based on the frequency spectrum of spectral signature.The spectral signature of each pixel in the hyperspectral image is taken as a discrete signal,and the frequency spectrum is obtained using discrete Fourier transform.The discrepancy of frequency spectrum between ground objects' spectral signatures is visible,thus the difference between frequency spectra of reference and target spectral signature is used to measure the spectral similarity.Canberra distance is introduced to increase the contribution from higher frequency components.Then,the number of harmonics involved in the proposed algorithm is determined after analyzing the frequency spectrum energy cumulative distribution function of ground object.In order to evaluate the performance of the proposed algorithm,two hyperspectral remote sensing images are adopted as experimental data.The proposed algorithm is compared with spectral angle mapper (SAM),spectral information divergence (SID) and Euclidean distance (ED) using the product accuracy,user accuracy,overall accuracy,average accuracy and Kappa coefficient.The results show that the proposed algorithm can be applied to hyperspectral image classification effectively.

The quantitative evaluation of application of hyperspectral data based on multi-parameters joint optimization

In order to evaluate the mineral identification of the hyperspectral data and make a trade-off of the imaging system parameters,a quantitative evaluation approach based on the multi-parameters joint optimization is proposed for the hyperspectral remote sensing.In the proposed approach,the mineral identification is defined as the number of the minerals identified and the key imaging parameters employed include ground sample distance(GSD)and spectral resolution(SR).Certain limitations are found among parameters that are used for analyzing the imaging processes.The constraints include the industrial manufacturing level,application requirements and the quantitative relationship among the GSD,the SR and the signal-to-noise ratio(SNR).Regression analysis is used to investigate the quantitative relationship between the mineral identification and the key imaging system parameters.Then,an optimization model for the trade-off study is established by combining the regression equation with the constraints.The airborne hyperspectral image collected by Hymap is applied to evaluate the performance of the proposed approach.The experimental results reveal that the approach can achieve the evaluation of the mineral identification and the trade-off of key imaging system parameters.The error of the prediction is within one kind of mineral.

The spatial scaling effect of the discrete-canopy effective leaf area index retrieved by remote sensing

The leaf area index(LAI) is a critical biophysical variable that describes canopy geometric structures and growth conditions.It is also an important input parameter for climate,energy and carbon cycle models.The scaling effect of the LAI has always been of concern.Considering the effects of the clumping indices on the BRDF models of discrete canopies,an effective LAI is defined.The effective LAI has the same function of describing the leaf density as does the traditional LAI.Therefore,our study was based on the effective LAI.The spatial scaling effect of discrete canopies significantly differed from that of continuous canopies.Based on the directional second-derivative method of effective LAI retrieval,the mechanism responsible for the spatial scaling effect of the discrete-canopy LAI is discussed and a scaling transformation formula for the effective LAI is suggested in this paper.Theoretical analysis shows that the mean values of effective LAIs retrieved from high-resolution pixels were always equal to or larger than the effective LAIs retrieved from corresponding coarse-resolution pixels.Both the conclusions and the scaling transformation formula were validated with airborne hyperspectral remote sensing imagery obtained in Huailai County,Zhangjiakou,Hebei Province,China.The scaling transformation formula agreed well with the effective LAI retrieved from hyperspectral remote sensing imagery.

A Fast Clonal Selection Algorithm for Feature Selection in Hyperspectral Imagery

Clonal selection feature selection algorithm （CSFS） based on clonal selection algorithm （CSA）, a new computational intelligence approach, has been proposed to perform the task of dimensionality reduction in high-dimensional images, and has better performance than traditional feature selection algorithms with more computational costs. In this paper, a fast clonal selection feature selection algorithm （FCSFS） for hyperspectral imagery is proposed to improve the convergence rate by using Cauchy mutation instead of non-uniform mutation as the primary immune operator. Two experiments are performed to evaluate the performance of the proposed algorithm in comparison with CSFS using hyperspectral remote sensing imagery acquired by the pushbroom hyperspectral imager （PHI） and the airborne visible/infrared imaging spectrometer （AVlRIS）, respectively. Experimental results demonstrate that the FCSFS converges faster than CSFS, hence providing an effective new option for dimensionality reduction of hyperspectral remote sensing imagery.