About feature matching and the match table

The Generate Rubbersheet Links, Transfer Attributes, and Detect Feature Changes tools use feature matching techniques to identify corresponding features and produce various results. This document describes the feature matching process and explains feature matching information found in the match table.

What is feature matching?

Feature matching means finding corresponding features from two similar datasets based on a search distance. One of the datasets is named source and the other target, especially when the feature matching is used to derive rubbersheet links or to transfer attributes from source to target data. These datasets overlap each other but are not perfectly aligned due to inconsistent data collection, changes over time, or other reasons. Figure 1 shows an example of streets, where the source features come from a commercial data provider and the target features are built and maintained by a city government.

Illustration of similar but inconsistent datasets for feature matching
Figure 1: Illustration of similar but inconsistent datasets for feature matching

The feature matching process analyzes the source and target topology, detects certain feature patterns, matches the patterns, and matches features within the patterns. The accuracy of feature matching depends on data similarity, complexity, and quality. In general, the more similar the two datasets, the better matching results. Normally, a high percentage of successful matching can be achieved, while uncertainty and errors may occur and require postinspection and corrections.

Feature attributes can optionally help determine the right match in feature matching. If one or more pairs of match fields are specified, spatially matched features are checked against the match fields. For example, if one source feature spatially matches two candidate target features, but one of the target features has matching attribute values and the other doesn't, then the former is chosen as the final match. The condition of attribute match affects the level of confidence of the feature matching.

What is in the match table?

The Generate Rubbersheet Links, Transfer Attributes, and Detect Feature Changes tools can optionally produce a match table. The match table provides complete feature matching information with the following fields to help understand the result and to facilitate postinspection and further analysis:

  • SRC_FID—The source feature ID. The value is -1 for an unmatched source feature.
  • TGT_FID—The target feature ID. The value is -1 for an unmatched target feature.
  • FM_GRP—The unique group ID for matched features and -1 for unmatched source or target features
  • FM_MN—The match relationships between source and target features in the format of m:n, where m and n are the number of source features and target features in a match group, respectively. For example, 1:1 is a one-to-one match; and 3:2 a three-to-two match. For unmatched source or target features, the value in this field is N/A for not applicable.
  • FM_CONF—This field stores the following values representing the level of confidence of the feature matching:
    • 100 - match confirmed by the specified match fields
    • 75 - match not confirmed by the specified match fields, either because match fields are not specified or no values are found in the specified match fields
    • 50 - match with a field value difference in the specified match fields
    • 0 - unmatched source or target features

The next section has example feature matching scenarios and the matching information in the match tables. For simplicity, attribute match is assumed for all matched features, reflected in FM_CONF value of 100.

Match groups and match relationships

Since the feature matching is based on feature topology and spatial patterns, where one or more source features and one or more target features are recognized as having a matching topological structure or spatial pattern, they become a match group. Within each match group, the match relationship is defined by the number of source features (m) versus the number of target features (n), as explained below and illustrated in Figure 2:

  • One-to-one (1:1) match

    One source feature matches one target feature; they belong to the same match group.

  • One-to-many (1:m) match

    One source feature matches multiple target features; all of them belong to the same match group.

  • Many-to-one (m:1) match

    Multiple source features match one target feature; all of them belong to the same match group.

  • Many-to-many (m:n) match

    Multiple source features match multiple target features; all of them belong to the same match group.

Illustrations of match information
Figure 2: Illustrations of match information