Land cover classification by an artificial neural network with ancillary information

Abstract

Remote sensing is an important source of land cover data required by many GIS users. Land cover data are typically derived from remotely–sensed data through the application of a conventional statistical classification. Such classification techniques are not, however, always appropriate, particularly as they may make untenable assumptions about the data and their output is hard, comprising only the code of the most likely class of membership. Whilst some deviation from the assumptions may be tolerated and a fuzzy output may be derived, making more information on class membership properties available, alternative classification procedures are sometimes required. Artificial neural networks are an attractive alternative to the statistical classifiers and here one is used to derive a fuzzy classification output from a remotely–sensed data set that may be post–processed with ancillary data available in a GIS to increase the accuracy with which land cover may be mapped. With the aid ancillary information on soil type and prior knowledge of class occurrence the accuracy of an artificial neural network classification was increased by 29–93 to 77–37 per cent. An artificial neural network can therefore be used generate a fuzzy classification output that may be used with other data sets in a GIS, which may not have been available to the producer of the classification, to increase the accuracy with which land cover may be classified.     

基于蚁群规则挖掘算法的多特征遥感数据分类

蚁群算法作为一种新型的智能优化算法,已经成功应用在许多领域,然而应用蚁群优化算法进行遥感数据处理则是一个新的研究热点。蚁群规则挖掘算法是基于分类规则挖掘进行分类,能够处理多特征的数据。因此,论文将蚁群规则挖掘算法应用到多特征遥感数据分类处理中,并采用北京地区的Landsat TM和 Envisat ASAR数据作为实验数据,对选择的遥感数据进行了多特征分类实验。实验结果分别与最大似然分类法、C4.5方法进行对比,分析表明:1)蚁群规则挖掘算法是一种无参数分类的智能方法,具有很好的鲁棒性,2)能够挖掘较简单的分类规则;3)能够充分利用多源遥感数据等。它可以充分利用多特征数据进行土地覆盖分类,从而能够提高分类的效率。     

Mining significant crisp-fuzzy spatial association rules

Spatial association rule mining (SARM) is an important data mining task for understanding implicit and sophisticated interactions in spatial data. The usefulness of SARM results, represented as sets of rules, depends on their reliability: the abundance of rules, control over the risk of spurious rules, and accuracy of rule interestingness measure (RIM) values. This study presents crisp-fuzzy SARM, a novel SARM method that can enhance the reliability of resultant rules. The method firstly prunes dubious rules using statistically sound tests and crisp supports for the patterns involved, and then evaluates RIMs of accepted rules using fuzzy supports. For the RIM evaluation stage, the study also proposes a Gaussian-curve-based fuzzy data discretization model for SARM with improved design for spatial semantics. The proposed techniques were evaluated by both synthetic and real-world data. The synthetic data was generated with predesigned rules and RIM values, thus the reliability of SARM results could be confidently and quantitatively evaluated. The proposed techniques showed high efficacy in enhancing the reliability of SARM results in all three aspects. The abundance of resultant rules was improved by 50% or more compared with using conventional fuzzy SARM. Minimal risk of spurious rules was guaranteed by statistically sound tests. The probability that the entire result contained any spurious rules was below 1%. The RIM values also avoided large positive errors committed by crisp SARM, which typically exceeded 50% for representative RIMs. The real-world case study on New York City points of interest reconfirms the improved reliability of crisp-fuzzy SARM results, and demonstrates that such improvement is critical for practical spatial data analytics and decision support.     

A rough set approach to the discovery of classification rules in spatial data

This paper proposes a novel rough set approach to discover classification rules in real‐valued spatial data in general and remotely sensed data in particular. A knowledge induction process is formulated to select optimal decision rules with a minimal set of features necessary and sufficient for a remote sensing classification task. The approach first converts a real‐valued or integer‐valued decision system into an interval‐valued information system. A knowledge induction procedure is then formulated to discover all classification rules hidden in the information system. Two real‐life applications are made to verify and substantiate the conceptual arguments. It demonstrates that the proposed approach can effectively discover in remotely sensed data the optimal spectral bands and optimal rule set for a classification task. It is also capable of unraveling critical spectral band(s) discerning certain classes. The framework paves the road for data mining in mixed spatial databases consisting of qualitative and quantitative data.     

空间数据挖掘技术研究进展

空间数据具有海量、非线性、多尺度、高维和模糊性等复杂性特点,空间数据挖掘技术是对空间数据中非显性的知识、空间关系等模式的自动提取。该文从空间数据挖掘的知识类型、方法、体系结构、过程以及与GIS系统集成等方面对其进行综述。重点阐述空间特征及区分规则、空间分类及聚类规则、空间分布及关联规则、空间序列及演化规则等知识类型以及统计分析、机器学习、探索性数据分析、可视化分析等数据挖掘方法。通过对空间数据挖掘理论、应用和系统实现等方面研究方向、存在问题的分析,指出集数据库、知识库、专家系统、决策支持系统、可视化工具、网络等技术于一体的空间数据挖掘系统是其主要发展方向。     

Using rough set theory to identify villages affected by birth defects: the example of Heshun,Shanxi, China

This article uses rough set theory to explore spatial decision rules in neural-tube birth defects and searches for novel spatial factors related to the disease. The whole rule induction process includes data transformation, searching for attribute reducts, rule generation, prediction or classification, and accuracy assessment. We use Heshun as an example, where neural-tube birth defects are prevalent, to validate the approach. About 50% of the villages in Heshun are used as the sample data, from which all of the rules are extracted. Meanwhile, the other villages are used as reference data. The rules extracted from the training data are then applied to the reference data. The result shows that the rules' generalization is reasonably good. Moreover, a novel relationship between the spatial attributes and the neural-tube birth defects was discovered. That is, the villages that lie in Watershed 9 of this district and that are also associated with a gradient of between 16° and 25° are vulnerable to neural-tube birth defects. This result paves the road for predicting where high rates of neural-tube birth defects will occur and can be used as a preliminary step in finding a direct cause for the disease.     

基于自适应模糊神经网络预测金属——EDTA配合物稳定常数

采用自适应模糊神经网络的方法,以金属离子外层主量子数(n)、电荷(Z)、半径(r)、适配价轨道数因子(w)及价电子结构因子(S)等为参数,关联金属—EDTA配合物稳定常数。利用减法聚类算法确定模糊神经网络的结构,并结合模糊推理系统进行该网络参数的调整,网络仿真的结果是满意的。在此基础上,预测了13种金属—EDTA配合物稳定常数。     

空间关联规则在土地利用与地形特征关系研究中的应用

空间关联规则是空间数据挖掘的重要内容,其结果表明了各种空间对象之间的关联关系．本研究以福州地区作为试验区,以DEM、坡度、坡向等地形特征以及2009年福州地区土地利用现状作为基础数据,利用Apriori算法从中提取出地形特征与土地利用现状之间的关联关系,讨论并分析两者之间关联规则的提取结果及空间关联规则提取方法的优缺点;研究结果表明了2009年福州地区的土地利用现状分布,即林地多,耕地、住宅用地等偏少的情况,林地分布在各种地形上且与坡向之间无强关联性;而且对于不同的最小置信度和支持度,该算法所提取的结果有所不同,如何提高算法效率、合理的设置最小置信度和支持度以及提取结果的评价与解释等将是今后进一步研究的重点．     

空间关联规则在土地利用与地形特征关系研究中的应用

空间关联规则是空间数据挖掘的重要内容,其结果表明了各种空间对象之间的关联关系．本研究以福州地区作为试验区,以DEM、坡度、坡向等地形特征以及2009年福州地区土地利用现状作为基础数据,利用Apriori算法从中提取出地形特征与土地利用现状之间的关联关系,讨论并分析两者之间关联规则的提取结果及空间关联规则提取方法的优缺点...     

A Comparison of the Weights-of-Evidence Method and Probabilistic Neural Networks

The need to integrate large quantities of digital geoscience information to classify locations as mineral deposits or nondeposits has been met by the weights-of-evidence method in many situations. Widespread selection of this method may be more the result of its ease of use and interpretation rather than comparisons with alternative methods. A comparison of the weights-of-evidence method to probabilistic neural networks is performed here with data from Chisel Lake-Andeson Lake, Manitoba, Canada. Each method is designed to estimate the probability of belonging to learned classes where the estimated probabilities are used to classify the unknowns. Using these data, significantly lower classification error rates were observed for the neural network, not only when test and training data were the same (0.02 versus 23%), but also when validation data, not used in any training, were used to test the efficiency of classification (0.7 versus 17%). Despite these data containing too few deposits, these tests of this set of data demonstrate the neural network's ability at making unbiased probability estimates and lower error rates when measured by number of polygons or by the area of land misclassified. For both methods, independent validation tests are required to ensure that estimates are representative of real-world results. Results from the weights-of-evidence method demonstrate a strong bias where most errors are barren areas misclassified as deposits. The weights-of-evidence method is based on Bayes rule, which requires independent variables in order to make unbiased estimates. The chi-square test for independence indicates no significant correlations among the variables in the Chisel Lake–Andeson Lake data. However, the expected number of deposits test clearly demonstrates that these data violate the independence assumption. Other, independent simulations with three variables show that using variables with correlations of 1.0 can double the expected number of deposits as can correlations of –1.0. Studies done in the 1970s on methods that use Bayes rule show that moderate correlations among attributes seriously affect estimates and even small correlations lead to increases in misclassifications. Adverse effects have been observed with small to moderate correlations when only six to eight variables were used. Consistent evidence of upward biased probability estimates from multivariate methods founded on Bayes rule must be of considerable concern to institutions and governmental agencies where unbiased estimates are required. In addition to increasing the misclassification rate, biased probability estimates make classification into deposit and nondeposit classes an arbitrary subjective decision. The probabilistic neural network has no problem dealing with correlated variables—its performance depends strongly on having a thoroughly representative training set. Probabilistic neural networks or logistic regression should receive serious consideration where unbiased estimates are required. The weights-of-evidence method would serve to estimate thresholds between anomalies and background and for exploratory data analysis.     

具有逆辨识结构的模糊神经网络控制及其应用

该文提出一种新型模糊神经网络结构及算法。在这种控制方案中,采用三层模糊神经网络控制器和神经网络逆辨识控制器相结合的结构。计算机仿真研究和实际应用表明,采用新型模糊神经网络控制方法,对大滞后非线性系统的控制是有效的。     

PROGRAM FOR PROCESSING OF HISTORICAL LABORATORY DATA BASED ON FUZZY SET THEORY AS A TOOL FOR DECISION SUPPORT

A computer program(FEA)is presented for processing historical laboratory data.It performs on a list ofsample entries stored in a laboratory information management system.Using an algorithm which is basedon fuzzy set theory,FEA classifies the entries into a limited number of clusters called sample types.Theclassification is fully user-defined.The program transforms the historical data into a representation whichis more suitable for studying the performance of the laboratory or which can be used as preparation for asimulation project.     

地学应用中的遥感图像处理若干问题的分析

遥感技术在地理学应用中是如何从遥感影像上直观、准确的得到所需的信息,为本专业服务。文章从地学应用部门在进行遥感影像处理时遇到的几个问题:多光谱数据的选取与合成;多源信息的复合;新型图像分类器的应用;专题提取的精度等方面的进展作了分析。     

Interactive and visual fuzzy classification of remotely sensed imagery for exploration of uncertainty

In this study, we propose, describe, and demonstrate a new geovisualization tool to demonstrate the use of exploratory and interactive visualization techniques for a visual fuzzy classification of remotely sensed imagery. The proposed tool uses dynamically linked views, consisting of an image display, a parallel coordinate plot, a 3D feature space plot, and a classified map with an uncertainty map. It allows a geoscientist to interact with the parameters of a fuzzy classification algorithm by visually adjusting fuzzy membership functions and fuzzy transition zones of land-cover classes. The purpose of this tool is to improve insight into fuzzy classification of remotely sensed imagery and related uncertainty. We tested our tool with a visual fuzzy land-cover classification of a Landsat 7 ETM+ image of an area in southern France characterized by objects with indeterminate boundaries. Good results were obtained with the visual classifier. Additionally, a focus-group user test of the tool showed that insight into a fuzzy classification algorithm and classification uncertainty improved considerably.     

制图数据的模糊分级处理模型研究

传统的制图数据分级方法存在对原始数据信息的歪曲、普适性不强及计算复杂等问题。基于此,结合现实分级问题的模糊性,提出基于模糊统计分析模型的制图数据分级处理方法。首先通过专家系统获取各模糊样本集,利用统计分析方法求得样本分布函数;然后利用分布函数获得模糊隶属函数,进而求取各模糊集的最模糊点;最后根据最模糊点获得各模糊集的区域划分,从而实现对制图数据的分级处理。该方法不需要对影响级别划分的多因子进行分析和转换,降低了计算的复杂度;另外,该方法是在获得原始数据实际分布的基础上进行的,在后续的分级过程中避免了对原始数据信息的歪曲。     

Direct induction of compact rule-based classifiers for resource mapping

Abstract

Rule-based classifiers are used regularly with geographical information systems to map categorical attributes on the basis of a set of numeric or unordered categorical attributes. Although a variety of methods exist for inducing rule-based classifiers from training data, these tend to produce large numbers of rules when the data has noise. This paper describes a method for inducing compact rule-sets whose classification accuracy can, at least in some domains, compare favourably with that achieved by larger less succinct rule-sets produced by alternative methods. One rule is induced for each output class. The condition list for this rule represents a box in n-dimensional attribute space, formed by intersecting conditions which exclude other classes. Despite this simplicity, the classifier performed well in the test application prediction of soil classes in the Port Hills, New Zealand, on the basis of regolith type and topographic attributes obtained from a digital terrain model.     

SHORT COMMUNICATION CLASSIFICATION METHODS FOR MULTIVARIATE QUALITY PARAMETERS

There are many chemical products where product conformity is decided upon by qualitative humanjudgements of overall product quality.Nowadays,quantitative instrumentally determined qualityparameters become available which are intended to replace such qualitative judgements by means ofautomatic decision rules using multivariate specification limits.Six classification methods to derive suchlimits are compared in terms of their power to predict corresponding human judgements on overall colorconformity of 17 dyestuffs based on historical quality data.Standard statistical classification methodsturned out to be unacceptable for the routine generation of decision rules because of the frequent distinctsuboptimality of their predictive power.Instead,a simple non-statistical classification method utilizinga priori knowledge about the underlying data structure yielded uniformly satisfactory decision rules.     

Extensions of GAP‐tree and its implementation based on a non‐topological data model

This paper discusses extensions of GAP‐trees from three aspects and its implementation based on non‐topological structure in order to enhance access to large vector data sets. First of all, we apply cartographic generalization rules to build a generalization procedure of the GAP‐tree, which makes coarse representations more consistent with human cognition. Second, we replace the three‐dimensional (pseudo‐) Reactive‐tree index with a 2D R‐tree index and a B‐tree index to improve the system efficiency. Finally, we compress a binary GAP‐tree into multi‐way GAP‐trees in order to reduce data redundancy. The shallower multi‐way GAP‐trees not only eliminate redundant data but also accelerate the system's response time. The extensions have been successfully implemented in PostgreSQL. A test of Beijing's land‐use data at the 1:10 000 scale demonstrates that the extended GAP‐trees are efficient, compact, and easy to implement.     

An enhanced supervised spatial decision support system of image classification: consideration on the ancillary information of paddy rice area

The analysis, measurement, and computation of remote sensing images often require an enhanced supervised classification technique to develop an efficient spatial decision support system. Rice is a crop of global importance, which has drawn a great interest in using remote sensing techniques for evaluating its production. Ancillary information is widely used to improve the classification accuracy of satellite images. However, few of these studies questioned the importance and strategies of using this ancillary information. The enhanced decision support system in our study has two stages. In the first stage, the images are obtained from the remote sensing technique and the ancillary information is employed to increase the accuracy of classification. In the second stage, it is decided to construct an efficiently supervised classifier, which is used to evaluate the ancillary information. Back-propagation neural network (BPN) with extended delta bar delta (EDBD) algorithm is incorporated into our decision support classifier system. This classifier renders two crucial contributions: (1) the EDBD algorithm accelerates the convergence speed of the learning process and (2) the relative importance (RI) on each band of ancillary information is evaluated rationally.