A comparison of texture measures for the per-field classification of Mediterranean land cover

Land cover of a Mediterranean region was classified within an artificial neural network (ANN) on a per-field basis using Landsat Thematic Mapper (TM) imagery. In addition to spectral information, the classifier used geostatistical structure functions and texture measures extracted from the co-occurrence matrix. Geostatistical measures of texture resulted in a more accurate classification of Mediterranean land cover than statistics derived from the co-occurrence matrix. The primary advantage of geostatistical measures was their robustness over a wide range of land cover types, field sizes and forms of class mixing. Spectral information and the variogram (geostatistical texture measure) resulted in the highest overall classification accuracies.     

Efficiency and accuracy of per-field classification for operational crop mapping

A crop map of The Netherlands was created using a methodology that integrates multi-temporal and multi-sensor satellite imagery, statistical data on crop area and parcel boundaries from a 1?:?10?000 digital topographic map. In the first phase a crop field database was created by extracting static parcel boundaries from the digital topographic map and by adding dynamic crop boundaries using on-screen digitizing. In the next phase the crop type was determined from the spectral and phenological properties of each field. The resulting crop map has an accuracy larger than 80% for most individual crops and an overall accuracy of 90%. By comparing cost and man-hours it was demonstrated that per-field classification is more efficient than per-pixel classification and decreased the effort for classification from 1500 to 500 man-hours, but the effort for creating the crop field database was estimated at 2300 man-hours. The use of image segmentation techniques for deriving the crop field database was discussed. It was concluded that image segmentation cannot replace the use of a large-scale topographic map but, in the future, image segmentation may be used to map the dynamic crop boundaries within the topographic parcels.     

Advances in classification for land cover mapping using SPOT HRV imagery

Abstract

High-resolution data from the HRV (High Resolution Visible) sensors onboard the SPOT-1 satellite have been utilized for mapping semi-natural and agricultural land cover using automated digital image classification algorithms. Two methods for improving classification performance are discussed. The first technique involves the use of digital terrain information to reduce the effects of topography on spectral information while the second technique involves the classification of land-cover types using training data derived from spectral feature space. Test areas in Snowdonia and the Somerset Levels were used to evaluate the methodology and promising results were achieved. However, the low classification accuracies obtained suggest that spectral classification alone is not a suitable tool to use in the mapping of semi-natural cover types.     

Comparison of pixel- and object-based classification in land cover change mapping

Land use/land cover (LULC) change occurs when humans alter the landscape, and this leads to increasing loss, fragmentation and spatial simplification of habitat. Many fields of study require monitoring of LULC change at a variety of scales. LULC change assessment is dependent upon high-quality input data, most often from remote sensing-derived products such as thematic maps. This research compares pixel- and object-based classifications of Landsat Thematic Mapper (TM) data for mapping and analysis of LULC change in the mixed land use region of eastern Ontario for the period 1995–2005. For single date thematic maps of 10 LULC classes, quantitative and visual analyses showed no significant accuracy difference between the two methods. The object-based method produced thematic maps with more uniform and meaningful LULC objects, but it suffered from absorption of small rare classes into larger objects and the incapability of spatial parameters (e.g. object shape) to contribute to class discrimination. Despite the similar map accuracies produced by the two methods, temporal change maps produced using post-classification comparison (PCC) and analysed using intensive visual analysis of errors of omission and commission revealed that the object-based maps depicted change more accurately than maximum likelihood classification (MLC)-derived change maps.     

Sub-pixel mapping of urban land cover using multiple endmember spectral mixture analysis: Manaus, Brazil

The spatial and spectral variability of urban environments present fundamental challenges to deriving accurate remote sensing products for urban areas. Multiple endmember spectral mixture analysis (MESMA) is a technique that potentially addresses both challenges. MESMA models spectra as the linear sum of spectrally pure endmembers that vary on a per-pixel basis. Spatial variability is addressed by mapping sub-pixel components of land cover as a combination of endmembers. Spectral variability is addressed by allowing the number and type of endmembers to vary from pixel to pixel. This paper presents an application of MESMA to map the physical components of urban land cover for the city of Manaus, Brazil, using Landsat Enhanced Thematic Mapper (ETM+) imagery.We present a methodology to build a regionally specific spectral library of urban materials based on generalized categories of urban land-cover components: vegetation, impervious surfaces, soil, and water. Using this library, we applied MESMA to generate a total of 1137 two-, three-, and four-endmember models for each pixel; the model with the lowest root-mean-squared (RMS) error and lowest complexity was selected on a per-pixel basis. Almost 97% of the pixels within the image were modeled within the 2.5% RMS error constraint. The modeled fractions were used to generate continuous maps of the per-pixel abundance of each generalized land-cover component. We provide an example to demonstrate that land-cover components have the potential to characterize trajectories of physical landscape change as urban neighborhoods develop through time. Accuracy of land-cover fractions was assessed using high-resolution, geocoded images mosaicked from digital aerial videography. Modeled vegetation and impervious fractions corresponded well with the reference fractions. Modeled soil fractions did not correspond as closely with the reference fractions, in part due to limitations of the reference data. This work demonstrates the potential of moderate-resolution, multispectral imagery to map and monitor the evolution of the physical urban environment.     

AVHRR channel selection for land cover classification

Mapping land cover of large regions often requires processing of satellite images collected from several time periods at many spectral wavelength channels. However, manipulating and processing large amounts of image data increases the complexity and time, and hence the cost, that it takes to produce a land cover map. Very few studies have evaluated the importance of individual Advanced Very High Resolution Radiometer (AVHRR) channels for discriminating cover types, especially the thermal channels (channels 3, 4 and 5). Studies rarely perform a multi-year analysis to determine the impact of inter-annual variability on the classification results. We evaluated 5 years of AVHRR data using combinations of the original AVHRR spectral channels (1-5) to determine which channels are most important for cover type discrimination, yet stabilize inter-annual variability. Particular attention was placed on the channels in the thermal portion of the spectrum. Fourteen cover types over the entire state of Colorado were evaluated using a supervised classification approach on all two-, three-, four- and five-channel combinations for seven AVHRR biweekly composite datasets covering the entire growing season for each of 5 years. Results show that all three of the major portions of the electromagnetic spectrum represented by the AVHRR sensor are required to discriminate cover types effectively and stabilize inter-annual variability. Of the two-channel combinations, channels 1 (red visible) and 2 (near-infrared) had, by far, the highest average overall accuracy (72.2%), yet the inter-annual classification accuracies were highly variable. Including a thermal channel (channel 4) significantly increased the average overall classification accuracy by 5.5% and stabilized interannual variability. Each of the thermal channels gave similar classification accuracies; however, because of the problems in consistently interpreting channel 3 data, either channel 4 or 5 was found to be a more appropriate choice. Substituting the thermal channel with a single elevation layer resulted in equivalent classification accuracies and inter-annual variability.     

M5 model tree for land cover classification

Tree based regression models like a M5 algorithm represent a promising development in machine learning research. A recent study suggests that a M5 model tree algorithm can be used for classification problems after some modification. This letter explores the usefulness of a M5 model tree for classification problems using multispectral (Landsat‐7 Enhanced Thematic Mapper Plus (ETM+)) for a test area in eastern England. Classification accuracy achieved by using a M5 model tree is compared with a univariate decision tree with and without using boosting. Results show that the M5 model tree achieves a significantly higher level of classification accuracy than a decision tree and works equally well to a boosted decision tree. Further, a model tree based classification algorithm works well with small as well as noisy datasets.     

AVHRR composite period selection for land cover classification

Multitemporal satellite image datasets provide valuable information on the phenological characteristics of vegetation, thereby significantly increasing the accuracy of cover type classifications compared to single date classifications. However, the processing of these datasets can become very complex when dealing with multitemporal data combined with multispectral data. Advanced Very High Resolution Radiometer (AVHRR) biweekly composite data are commonly used to classify land cover over large regions. Selecting a subset of these biweekly composite periods may be required to reduce the complexity and cost of land cover mapping. The objective of our research was to evaluate the effect of reducing the number of composite periods and altering the spacing of those composite periods on classification accuracy. Because inter-annual variability can have a major impact on classification results, 5 years of AVHRR data were evaluated. AVHRR biweekly composite images for spectral channels 1-4 (visible, nearinfrared and two thermal bands) covering the entire growing season were used to classify 14 cover types over the entire state of Colorado for each of five different years. A supervised classification method was applied to maintain consistent procedures for each case tested. Results indicate that the number of composite periods can be halved-reduced from 14 composite dates to seven composite dates-without significantly reducing overall classification accuracy (80.4% Kappa accuracy for the 14-composite dataset as compared to 80.0% for a seven-composite dataset). At least seven composite periods were required to ensure the classification accuracy was not affected by inter-annual variability due to climate fluctuations. Concentrating more composites near the beginning and end of the growing season, as compared to using evenly spaced time periods, consistently produced slightly higher classification values over the 5 years tested (average Kappa of 80.3% for the heavy early/late case as compared to 79.0% for the alternate dataset case).     

A new histogram quantization algorithm for land cover mapping

Land cover mapping from multi-spectral satellite data is based primarily on spectral differences in land cover categories. Since only a limited number of cover types are desired in most cases, the images contain redundant information which unnecessarily complicates the digital mapping process. In this study, we have devised an algorithm to automatically and reproducibly quantize an image to be classified into a reduced number of digital levels, in most cases without a visually perceptible reduction in the image information content. The Flexible Histogram Quantization (FHQ) algorithm assumes that the histogram has one or two major peaks (representing water and/or land) and that most of the information of interest is in one peak. It aims to provide a sufficient quantization in the main peak of interest as well as in the tails of this peak by computing an optimized number of quantized levels and then identifying the range of digital values belonging to each level. A comparison of the FHQ with four existing quantization algorithms showed that the FHQ retained substantially more radiometric discrimination than histogram normalization, linear quantization, and scaling methods. Using a random sample of Landsat TM images and an AVHRR coverage of Canada, the average quantization error for the FHQ was 1.68 digital levels for an entire scene and 1.41 for land pixels only. Based on the 34 single-band test images included in the comparison, the radiometric resolution was reduced from 255 to 23.3 levels on the average, or by a factor of 10.94 for a multi-spectral image with n spectral bands. Compared to the other quantization methods, FHQ had a higher efficiency (by 65% to 148%), except for histogram equalization. FHQ also retained more information than histogram equalization (by 11%) but more importantly, it provided finer resolution in the tails of the main histogram peak (by 36-664%, depending on the position in the tails) for infrequent but potentially important land cover types. In addition, unlike the other methods the FHQ does not require a user-specified number of levels and therefore its results are fully reproducible. The F HQ can be used with single scenes, with radiometrically seamless mosaics, or when classifying radiometrically incompatible adjacent scenes. It is concluded that the FHQ provides an effective means for image quantization, as an automated pre-processing step in land cover mapping applications.     

Super-resolution land cover mapping with indicator geostatistics

Many satellite images have a coarser spatial resolution than the extent of land cover patterns on the ground, leading to mixed pixels whose composite spectral response consists of responses from multiple land cover classes. Spectral unmixing procedures only determine the fractions of such classes within a coarse pixel without locating them in space. Super-resolution or sub-pixel mapping aims at providing a fine resolution map of class labels, one that displays realistic spatial structure (without artifact discontinuities) and reproduces the coarse resolution fractions. In this paper, existing approaches for super-resolution mapping are placed within an inverse problem framework, and a geostatistical method is proposed for generating alternative synthetic land cover maps at the fine (target) spatial resolution; these super-resolution realizations are consistent with all the information available.More precisely, indicator coKriging is used to approximate the probability that a pixel at the fine spatial resolution belongs to a particular class, given the coarse resolution fractions and (if available) a sparse set of class labels at some informed fine pixels. Such Kriging-derived probabilities are used in sequential indicator simulation to generate synthetic maps of class labels at the fine resolution pixels. This non-iterative and fast simulation procedure yields alternative super-resolution land cover maps that reproduce: (i) the observed coarse fractions, (ii) the fine resolution class labels that might be available, and (iii) the prior structural information encapsulated in a set of indicator variogram models at the fine resolution. A case study is provided to illustrate the proposed methodology using Landsat TM data from SE China.     

Ensemble of support vector machines for land cover classification

This letter presents the results of two different ensemble approaches to increase the accuracy of land cover classification using support vector machines. Finite ensemble approaches, based on boosting and bagging and infinite ensemble created by embedding the infinite hypothesis in the kernel of support vector machines, are discussed. Results suggest that the infinite ensemble approach provides a significant increase in the classification accuracy in comparison to the radial basis function kernel‐based support vector machines. While using finite ensemble approaches, bagging works well and provides a comparable performance to the infinite ensemble approach, whereas boosting decreases the performance of support vector machines. Comparison in terms of computational cost suggests that finite ensemble approaches require a large processing time in comparison to the infinite ensemble approach.     

Object-based land cover classification using airborne LiDAR

Light Detection and Ranging (LiDAR) provides high resolution horizontal and vertical spatial point cloud data, and is increasingly being used in a number of applications and disciplines, which have concentrated on the exploit and manipulation of the data using mainly its three dimensional nature. LiDAR information potential is made even greater though, with its consideration of intensity.Elevation and intensity airborne LiDAR data are used in this study in order to classify forest and ground types quickly and efficiently without the need for manipulating multispectral image files, using a supervised object-orientated approach. LiDAR has the advantage of being able to create elevation surfaces that are in 3D, while also having information on LiDAR intensity values, thus it is a spatial and spectral segmentation tool. This classification method also uses point distribution frequency criteria to differentiate between land cover types. Classifications were performed using two methods, one that included the influence of the ground in heavily vegetated areas, and the other which eliminated the ground points before classification. The classification of three meanders of the Garonne and Allier rivers in France has demonstrated overall classification accuracies of 95% and 94% for the methods including and excluding the ground influence respectively. Five types of riparian forest were classified with accuracies between 66 and 98%. These forest types included planted and natural forest stands of different ages. Classifications of short vegetation and bare earth also produced high accuracies averaging above 90%.     

Fusion of multisensor multitemporal satellite data for land cover mapping

Efficient integration of remote sensing information with different temporal, spectral and spatial resolutions is important for accurate land cover mapping. A new temporal fusion classification (TFC) model is presented for land cover classification, based on statistical fusion of multitemporal satellite images. In the proposed model, the temporal dependence of multitemporal images is taken into account by estimating transition probabilities from the change pattern of a vegetation dynamics indicator (VDI). Extension of this model is applicable to Synthetic Aperture Radar (SAR) images and integration of multisensor multitemporal satellite images, concerning both temporal attributes and reliability of multiple data sources. The feasibility of the new method is verified using multitemporal Landsat Thematic Mapper (TM) and ERS SAR satellite images, and experimental results show improved performance over conventional methods.     

Boosting a fast neural network for supervised land cover classification

It is demonstrated that the use of an ensemble of neural networks for routine land cover classification of multispectral satellite data can lead to a significant improvement in classification accuracy. Specifically, the AdaBoost.M1 algorithm is applied to a sequence of three-layer, feed-forward neural networks. In order to overcome the drawback of long training time for each network in the ensemble, the networks are trained with an efficient Kalman filter algorithm. On the basis of statistical hypothesis tests, classification performance on multispectral imagery is compared with that of maximum likelihood and support vector machine classifiers. Good generalization accuracies are obtained with computation times of the order of 1 h or less. The algorithms involved are described in detail and a software implementation in the ENVI/IDL image analysis environment is provided.     

Automatic cost-effective method for land cover classification (ALCC)

The need for the detection and monitoring of changes in the environment is greater today than ever before. Through classification we can obtain insights into the state of the land surface. No known classification methods are fully automated, and their implementation requires preprocessing and postprocessing. This research provides a novel, fully automatic and cost-effective land cover classification method (ALCC). This novel automatic method does not require prior knowledge of the terrain or the assignment of training samples. The ALCC method is based on unsupervised classification methods, which is performed over the spectral indices rasters and six Landsat-8 30 m spatial resolution bands. The method was tested in three different study areas. Furthermore, all three study areas were classified by common supervised classification methods, namely, the Maximum Likelihood Classification (MLC) and the Random Forests (RF) method. For comparison accuracy, assessment of the three applied classification methods, namely, the figure of merit, overall agreement, omission and commission, were used. The results show that the overall agreement of the new automatic classification method for the Rijeka, Zagreb and Sarajevo study areas is 90.0%, 89.5% and 89.9%, respectively, and the overall agreement always falls between the overall agreement of the MLC method (88.1%, 88.9% and 86.7%, respectively) and the overall agreement of the RF method of classification (91.7%, 90.4% and 90.2%, respectively). These results confirm that this new automatic, cost-effective and accurate land cover classification method can be easily applied for numerous remote sensing applications.     

Examining region-based methods for land cover classification using stochastic distances

A recent alternative to standard pixel-based classification of remote-sensing data is region-based classification, which has proved to be particularly useful when analysing high-resolution imagery of complex environments, such as urban areas, or when addressing noisy data, such as synthetic aperture radar (SAR) images. First, following certain criteria, the imagery is decomposed into homogeneous regions, and then each region is classified into a class of interest. The usual method for region-based classification involves using stochastic distances, which measure the distances between the pixel distributions inside an unknown region and the representative distributions of each class. The class, which is at the minimum distance from the unknown region distribution, is assigned to the region and this procedure is termed stochastic minimum distance classification (SMDC). This study reports the use of methods derived from the original SMDC, Support Vector Machine (SVM), and graph theory, with the objective of identifying the most robust and accurate classification methods. The equivalent pixel-based versions of region-based analysed methods were included for comparison. A case study near the Tapajós National Forest, in Pará state, Brazil, was investigated using ALOS PALSAR data. This study showed that methods based on the nearest neighbour, derived from SMDC, and SVM, with a specific kernel function, are more accurate and robust than the other analysed methods for region-based classification. Furthermore, pixel-based methods are not indicated to perform the classification of images with a strong presence of noise, such as SAR images.     

Application of the MODIS global supervised classification model to vegetation and land cover mapping of Central America

While mapping vegetation and land cover using remotely sensed data has a rich history of application at local scales, it is only recently that the capability has evolved to allow the application of classification models at regional, continental and global scales. The development of a comprehensive training, testing and validation site network for the globe to support supervised and unsupervised classification models is fraught with problems imposed by scale, bioclimatic representativeness of the sites, availability of ancillary map and high spatial resolution remote sensing data, landscape heterogeneity, and vegetation variability. The System for Terrestrial Ecosystem Parameterization (STEP) - a model for characterizing site biophysical, vegetation and landscape parameters to be used for algorithm training and testing and validation - has been developed to support supervised land cover mapping. This system was applied in Central America using two classification systems based on 428 sites. The results indicate that: (1) it is possible to generate site data efficiently at the regional scale; (2) implementation of a supervised model using artificial neural network and decision tree classification algorithms is feasible at the regional level with classification accuracies of 75-88%; and (3) the STEP site parameter model is effective for generating multiple classification systems and thus supporting the development of global surface biophysical parameters.     

面向宏观地表分类的特征选择算法比较研究

为了成功将土地覆盖进行分类,选择合适的特征是至关重要的。针对利用MODIS数据进行宏观土地覆盖的分类问题,对三种典型的特征选择方法进行了比较研究。研究结果表明：分支定界法（BB）最适合于该土地覆盖分类问题,与此同时,ReliefF和mRMR方法在目标应用中的精度非常接近。研究结果同样表明进行特征选择是非常必要的,它不仅能够大大地降低计算复杂度,而且分类精度能够保持不变,甚至更高。     

30-second degree grid land cover classification of Asia

With the development of the global economy, environmental research has become more important than ever, especially in the Asian region. The objective of this study is to produce a land cover classification dataset for the whole of Asia using the NOAA AVHRR 1-km dataset. Ground data were mainly collected from existing thematic maps which were obtained from members of the Land Cover Working Group (LCWG) of the Asian Association of Remote Sensing (AARS). Classification was mainly based on cluster analysis of the monthly ratio of surface temperature and Normalized Difference Vegetation Index (NDVI) for seven months from April to October 1992. Additional variables, such as DEM, the maximum monthly composite NDVI in a year, and the minimum monthly composite NDVI in a year were also used in the classification processing. In order to add and improve ground data in the future, collected ground data will be published with the developed land cover dataset.     

Improving Thematic Mapper land cover classification using filtered data

In an attempt to alleviate the classification problems introduced by the higher spatial resolution of the Thematic Mapper in comparison to the Muitispectral Scanner, classifications were performed on two to six band combinations, first using Thematic Mapper bands only, and subsequently replacing band 5 by its mean-filtered and median-filtered counterpart. The combination of filtered data with non-filtered data smooths out scene noise while retaining some of the boundary detail.