WICS.xsd 2004-08-04 This XML Schema encodes the WICS operation request and response messages. Because different classifiers can involve very different parameters, both content and format, it is difficult to write a universal classification schema for all classifiers. In addition, the variable nature of the parameters will make it almost impossible to derive WSDL description of a universal classification service. Thus, a WICS server is restricted to provide only one specific classifier, and multiple classifiers shall be provided as separate WICS servers. This restriction simplifies significantly the definition of a WICS service. In this schema, an abstract, common definition for all WICS services is defined. When this schema is applied to a specific WICS server, the classifier-specific schema shall be defined and shall extend this common schema. The server shall then implement the abstract part of the schema. The abstract part of this schema is the wics:ClassifierParameters element. XML encoded operation request base, for all WICS operations except Get Capabilities. In this XML encoding, no "request" parameter is included, since the final element name will specify the specific operation. Service type identifier, where the string value is the OWS type abbreviation, namely "WICS". Specification version for OWS version and operation. See Version parameter Subclause 7.3.1 of OWS Common for more information. TBD. What is this? Each classifier shall define its own element to substitute for this abstract element. The parameters defined in this element shall include those specifying the structure of a classifier, such as the number of classes (TBR) and maximum iterations in an ISODATA classifier, or the numbers of layers and nodes in a multiple layer perceptron (MLP) neural network classifier. This element shall define only the static configuration parameters values of a classifier, and shall not include such dynamic values as the connection weights in a MLP network. Those latter parameters are obtained through the TrainClassifier operation and shall be included, together with the static configuration parameters, in a data file referenced by the TrainedParametersID. TBD. What is this? TBD. What is this? A brief description of the classifier, such as maximum likelihood supervised classification, ISODATA unsupervised classification etc. A URL point to a classifier-specific schema which defined the detail structure of the parameters used by the classifier. Unordered formats for the input images supported by the service. Unordered formats for the output images supported by the service. TBD. Is this list ordered or unordered? Class of a classifier (TBR). TBD. What is this? TBD. What is this? TBD. What is this? TBD. What is this? Request to a WICS to perform the GetCapabilities operation. This operation allows a client to retrieve a Capabilities XML document providing metadata for the specific WICS server. In this XML encoding, no "request" parameter is included, since the element name specifies the specific operation. XML encoded WICS GetCapabilities operation response. This document provides clients with service metadata about a specific service instance, including metadata about the tightly-coupled data served. If the server does not implement the updateSequence parameter, the server shall always return the complete Capabilities document, without the updateSequence parameter. When the server implements the updateSequence parameter and the GetCapabilities operation request included the updateSequence parameter with the current value, the server shall return this element with only the "version" and "updateSequence" attributes. Otherwise, all optional elements shall be included or not depending on the actual value of the Contents parameter in the GetCapabilities operation request. Contents section of WICS service metadata (or Capabilities) XML document. For the WICS, these contents are data and functions (TBR) that the WICS server supports (not data that can be accessed from this server). Request to a WICS to perform the GetClassification operation. This operation classifies an input SourceImage into a two-dimensional thematic data layer whose content are usually integer numerical values representing certain themes such as forest, grassland, and cropland. The SourceImage shall be an image grid coverage. In this XML encoding, no "request" parameter is included, since the element name specifies the specific operation. Reference to image to be classified (TBR). Two different types of input data are required by the GetClassification operation, in addition to the SourceImage. For an unsupervised classifier, WICS clients shall provide values for or use default values of the wics:ClassifierParameters element. For a supervised classifier, WICS clients shall provide a TrainedParametersID, as returned by TrainClassifier operation. Identification of format to be used for the classification result coverage (TBD). Response to a valid GetClassification operation request sent to a WICS. TBD. What is this? Request to a WICS to perform the TrainClassifier operation. This operation trains a supervised classifier with an input training data set whose target thematic meaning is already known. Each pixel in the input training data is assigned to one thematic class. The current classifier is trained using the training data to determine the parameters of the classifier. This operation only exists for a supervised classifier. The input training data used to train the specified classifier. The label (TBR) of the input training data. The input parameters for the selected classifier. These parameters shall determine the structure of a classifier but shall not determine the actual coefficients/weights of a classifier. For example, in a MLP network, the parameters shall determine the number of hidden layers and number of nodes in each hidden layer but they shall not determine the connection weights among layers. The latter shall be determined after the training is finished. TBD. When can this element be omitted? Response to a valid TrainClassifier operation request. Unique identifier that references a data file containing the trained parameters, to be used as an input in a GetClassification operation request sent to the same WICS server providing supervised classification. The trained parameters shall include the parameters necessary to perform the supervised classification using the same classifier as used in the training. Examples of the trained parameters include the mean and covariance matrices derived from a maximum likelihood classifier, and node connection weights of a multiple layer perceptron neural network classifier. The format of this parameter file is internal to a WICS server, and the client does not need to know this format. In a subsequent GetClassification request, WICS clients can send this parameter data file ID, together with the input image grid coverage to be classified, to the WICS server and the server shall get the correct parameters from the file referenced by the ID. Specifies the expiration data and time of the trained parameters kept by the server. After the expiration time, the trained parameters will be deleted, and the TrainedParameters ID that a client got from the TrainClassifier operation will become invalid. This element is included only for supervised classifiers. Request to a WICS to perform the DescribeClassifier operation. This operation allows a client to retrieve the description of the classifier. Response to a valid DescribeClassifier operation request sent to a WICS. It provides more detailed information about the classifier than the Capabilities, such as classification algorithm, keywords, and metadata. Reference to document that provides more detailed information about the classifier such as the algorithm description, algorithm theoretical document base (ATBD), classification schema in case of supervised classification. TBD. What is this used for? Why is this needed?