Er-Xue Chen

Chinese Academy of Forestry, Peping, Beijing, China

Are you Er-Xue Chen?

Claim your profile

Publications (9)1.36 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Studies are needed to evaluate the ability of Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) for forest aboveground biomass (AGB) extraction in mountainous areas. In this article, forest biomass was estimated at plot and stand levels, and different biomass grades, respectively. Light detection and ranging (LiDAR) data with about one hit per m2 were first used for forest biomass estimation at the plot level, with R 2 of 0.77. Then the LiDAR-derived biomass, as prior knowledge, was used to investigate the relationship between ALOS PALSAR data and biomass. The results showed that at each biomass level, the range of the back-scatter coefficient in HH and HV polarization (where H and V represent horizontal and vertical polarizations, respectively, and the first of the two letters refers to the transmission polarization and the second to the received polarization) was very large and there was no obvious relationship between the synthetic aperture radar (SAR) back-scatter coefficient and biomass at plot level. At stand level and in different biomass grades, the back-scatter coefficient increased with the increase of forest biomass, and a logarithm equation can be used to describe the relationship. The main reason may be that forest structure is complex at the plot level, while the average value could partly decrease the influence of forest structure at stand level. Meanwhile, terrain radiometric correction (TRC) was investigated and found effective for forest biomass estimation.
    International Journal of Remote Sensing 02/2012; 33(3):710-729. · 1.36 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In order to improve the land cover classification accuracy for SAR image, Support Vector Machine (SVM), which has wide applicability is used on the land cover classification of POLSAR image in this paper. The study site is located in Tahe County, Heilongjiang Province, China, and two scenes of quad-polarization Radarsat-2 SAR images were acquired. the land cover classification of single-temporal POLSAR image by SVM, and multi-temporal POLSAR image by SVM and maximum likelihood classification (MLC) is studied separately. Then all the classification results are evaluated. Some conclusions can be got according to the analysis of all results and accuracy: Firstly, it is difficult to distinguish the different types of vegetation for the similar scattering among them in July. However, water, whose scattering characteristic is simplex, can be distinguished from others easily. Scondly, in October, the scattering characteristics among forest, shrub, grass, crop are different, therefore it is easy to distinguish vegetation because of their one from others in this period. But for water, with reduced in winter, the river width narrows, compared with it in summer, water classification accuracy is lower in this period. Thirdly, joint July and October SAR data for classification, can offset espective their own disadvantages. and improve overall accuracy. And the last one, With the characteristics that different probability density distribution, small sample, non-linear and so on, SVM shows the wide applicability.
    Geoscience and Remote Sensing Symposium (IGARSS), 2012 IEEE International; 01/2012
  • [Show abstract] [Hide abstract]
    ABSTRACT: The main objective of this study was to investigate the potential of using Support Vector Machines (SVM) and Random forest (RF) to estimate forest above ground biomass (FAGB) by using multi-source remote sensing data. To do so, we introduced a basic flow of SVM to estimate FAGB from multisource remote sensing data. RF method was adept at identifying relevant features having main effects in multisource remote sensing data. Results show that: (i) In the stage of feature selection, the Random Forest model provide better results compared to the typical F-scores method. (ii) The optimal SVM model, based on the selection of features clearly demonstrate that the estimation accuracy increased by feature selection algorithm. (iii) Compared to the optimal KNN, BPNN and RBFNN model, the optimal SVM algorithm provided more accurate and robust result on the considered case.
    Geoscience and Remote Sensing Symposium (IGARSS), 2012 IEEE International; 01/2012
  • Ying Guo, Zeng-Yuan Li, Er-Xue Chen, Xu Zhang
    [Show abstract] [Hide abstract]
    ABSTRACT: In the recent years, the estimation of forest volume using radar data has developed greatly. However, as the radar data was large scale, the efficiency of processing based on KNN decreased seriously. Moreover, because the different K and distance measured method could result in the different accuracy, the treatment could have a low degree of automation under the condition of keeping the relatively better precision. Therefore, the study implemented a tool which could have the feature of fast and automatic processing radar data based on KNN. For enhancing the efficiency of processing, the tool was implemented in the way of parallelization by using the message passing interface (MPI) technology and run on the high performance cluster environment. To certain the suitable parameter automatically such as K and the appropriate distance measured method during the processing; the study used leave-one-out cross-validation method to check the precision and selected the optimum model based on the accuracy. The result shows that the tool accelerated the computation speed as eight time as before while ensuring the treatment precision and improved the automatic degree of the treatment. To some extend, it solved the bottleneck of processing large scale SAR data.
    Proc SPIE 06/2011;
  • [Show abstract] [Hide abstract]
    ABSTRACT: As forest biomass estimation depends on the various remote sensing factors, multiple regression model may not fully capture the complex relationship among the variable. Support Vector machines have already proven their ability in solving the nonlinear and multi-dimensional problems. This paper proposed to use Support Vector machines to improve the accuracy of forest biomass retrival with LiDAR and SPOT5 and adopted the leave-one-out method to validate the model accuracy. Results showed that (i) Support Vector machines had the best performance on the present data set as compared to the Back Propogation Networks,Radius Basis Function Networks and K nearest neighbor algorithm; (ii) compared to the single data source, the cooperative utilization of LiDAR and SPOT5 had the better result and this conclusion was suitable for the four using nonparametric methods; (iii) as the number of the input data dimension increasing, Support Vector machines was immune to the multi-dimension affection and performed better than other three schemes.
    01/2011;
  • Ying Guo, Zeng-Yuan Li, Er-Xue Chen, Xu Zhang
    [Show abstract] [Hide abstract]
    ABSTRACT: Forest type identification is one of most important contents of forest inventory. K nearest-neighbour algorithm has already proven their use in forest mapping. However, as the remote sensing data was large scale, the efficiency of processing based on KNN decreased seriously. Therefore, the study implemented a tool which could have the feature of fast processing multi-spectral data based on KNN. For enhancing the efficiency of processing, the tool was implemented in the way of parallelization by using the message passing interface (MPI) technology and run on the high performance cluster environment. By segmenting the input large scale image in some small block and parallel processing all these block, the computing time was shorten greatly. To certain the suitable parameter automatically such as K and the appropriate distance measured method during the processing, the study used leave-one-out cross validation method to check the precision and selected the optimum model based on the accuracy. The result shows that the tool accelerated the computation speed as eight time as before while ensuring the treatment precision and improved the automatic degree of the treatment. To some extend, it solved the bottleneck of processing large scale remote sensing data.
    Computer Science and Service System (CSSS), 2011 International Conference on; 01/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The ability to retrieve and monitor soil moisture and vegetation water content (VWC) is of great importance. Yet accurate retrieval of such information from microwave observations presents a big challenge, which calls for the development of high fidelity scattering models. In the literature, a ”discrete scatter” approach was usually deployed, which attempted to determine first the scattering behavior of the individual constituent of the canopy, then that of canopy as a whole by summing up either incoherently [1] [?] or coherently [2]-[3]. To simplify the problem, constituents of the canopy are modeled as canonical geometrical objects. For corn canopy, the stalks are modeled as dielectric circular cylinders with finite length, and the leaves are represented as thin dielectric disks with elliptic cross section. Since scattering from each of the canonical object serves as the base for further ”assembling”, it is expected to be accurately determined. However, mush is still desired in this regard. For a dielectric cylinder of finite length, in studying its scattering behavior the generalized Rayleigh-Gans approximation (GRGA) [4] is usually applied, which approximates the induced current in a finite cylinder by assuming infinite length. This method is valid for a needle shaped scatterer with radius much smaller than the wavelength. Yet caution must be taken even at L band when EM scattering from the stalk of a corn plant is to be evaluated using GRGA. It is also well known that GRGA fails to satisfy the reciprocity theorem [2]. In the evaluation of scattering amplitude of leaves, the GRGA method is usually used. However, caution must be taken here. At C band the wavelength is 5.6 cm, which is comparable to the length of minor axis of corn leaves, which presents an unfavorable condition in applying GRGA and thus appreciable error is expected in the predicted scattering amplitude. When corn canopy is at its early stage of growth, or when the- - incidence angle is not large, contribution from the underlying ground is appreciable and thus its accurate prediction is important. Yet this roughness effect has not been adequately addressed in canopy scattering models, where what is typically applied is conventional analytical method such as Kirchhoff approximation (KA), or the small perturbation method (SPM) [5], or the more advanced yet still improvement-needed integral equation method (IEM) [6]. In this study, we choose to apply a more rigorous treatment of the rough surface contribution using the recently advanced EAIEM model by the authors [7]. With the advancement of several scattering models of dielectric cylinder and disks and of rough surfaces, it is the aim of this paper to investigate if a coherent combination of these constituent models can improve predictive power of the resultant canopy scattering model. To be more specific, in analyzing electromagnetic scattering from a dielectric cylinder of finite length, we use the new approach that we have recently proposed [8], where a long cylinder is divided into a cluster of N identical sub-cylinder by using N - 1 hypothetic surfaces, for each the T matrix can be calculated stably in the numerical sense. The boundary conditions at the hypothetic interface are treated carefully. A system of equations is set up for each sub-cylinder, and the overall system of equations is coupled and linear, thus can be solved by appropriate iterative method. Moreover, the VPM method is found to be applicable to dielectric cylinders of arbitrary length as long as the T matrix is attainable for the elementary sub-cylinder. The applicable relative dielectric constant can go up to 70 (real part), which is normally the upper bound for corn stalks at C band. The radius of the cylinder can be as high as 5 wavelengths, a feature of the model that is expected to be useful for forest applications [9]. Scattering from rough surface is treated using the EAIEM model [7], which is a uni
    01/2011;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Fuzzy clustering algorithms have been successfully applied to POLSAR classification, but not to POLInSAR. In this paper, a Fuzzy C Means (FCM) clustering algorithm integrating the complementary physical information and statistical property contained in both polarimetric and interferometric data, is used for POLInSAR classification. At first, the area dominated by volume scattering is extracted from polarimetric information using unsupervised H-A-Alpha k-means Wishart classifier with the physical scattering mechanisms of different terrain types; and the volume scattering area (forest area) is further segmented in the feature space of the relative optimal interferometric coherence spectrum A1 and A2. Then a robust unsupervised fuzzy C means (FCM) classifier initialized with the results of the segmentation is applied to the polarimetric interferometric coherency data sets corresponding to the volume scattering area. This will not only take into account the scattering mechanisms of the data, so that the results of the classification have definite physical meaning, but also avoid the problem that the initial value of FCM algorithm is difficult to identify. The proposed method is evaluated and compared with k-means Wishart classifier using repeat pass E-SAR L band polarimetric interfer-ometric SAR data and the corresponding auxiliary image. Preliminary results show that the proposed method has better performance.
    01/2010;
  • [Show abstract] [Hide abstract]
    ABSTRACT: For technical and other reasons there is a dilemma that data providers cannot find an appropriate way to redistribute spatial forest data and data users who need spatial data cannot access and integrate available forest resources information. To overcome this dilemma, this paper proposed a spatial forest information system based on Web service using an open source software approach. With Web service based architecture, the system can enable interoperability, integrate Web services from other application servers, reuse codes, and shorten the development time and cost. At the same time, it is possible to extend the local system to a regional or national spatial forest information system. The growth of Open Source Software (OSS) provides an alternative choice to proprietary software for operating systems, web servers, Web-based GIS applications and database management systems. Using open source software to develop spatial forest information systems can greatly reduce the cost while providing high performance and sharing spatial forest information. We chose open source software to build a prototype system for Xixia County, Henan Province, China. By integrating OSS packages Deegree and UMN MapServer which are compliant to the OGC open specifications, the prototype system enables users to access spatial forest information and travelling information of Xixia County which come from two different data servers via a standard Web browser and promotes spatial forest information sharing.
    Journal of Forestry Research 18(2):85-90.