calcDiscrepancyMeasure_ED2: Calculate ED2 segmentation discrepancy measure

In joaofgoncalves/SegOptim: Optimization of Image Segmentation Parameters for Object-Based Image Analysis (OBIA)

Calculate ED2 segmentation discrepancy measure

Description

This function implements the ED2 discrepancy measure used to assess a given image segmentation solution. It is based on Liu et al., 2012 with some corrections introduced later by Novelli et al., 2017.

Usage

calcDiscrepancyMeasure_ED2(
  refData,
  segmData,
  thresh = 50,
  verbose = FALSE,
  ...
)

Arguments

refData	Reference polygons data. Either a string containing a path to a vector data set or a simple features geometry object (sf). See details below.
segmData	Segmentation polygons data. Either a string containing a path to a vector data set or a simple features geometry object (sf).
thresh	A threshold percentage used to select segments for each reference polygon. Calculated as: \%SegmIntersected = (AreaIntersected / AreaSegment) \times 100. If %SegmIntersected > thresh then that segment will be selected (default: 50).
verbose	Print progress messages? (default: FALSE)
...	Further arguments passed to read_sf function.

Details

A "reference polygon" can be considered a bounded geographic region (Castilla, 2003; Lang et al., 2010). A basic region is an indivisible unit with respect to certain perspectives of human cognition (Castilla and Hay, 2008) and can often be identified easily from its surroundings. In this case a reference polygon is considered a geographic object of interest. These reference polygons are typically sampled from EO imagery and delineated through manual interpretation or from field surveys. The size of a reference polygon is typically greater than that of a pixel in an image and smaller than the whole extent of that image.

Value

A numeric vector with several outputs:

• ED2 - The Euclidean Distance measure 2 (Liu et al., 2012);

• PSE - Potential Segmentation Error;

• NSR - Number-of-Segments Ratio;

• NrRefPolygons - Total number of reference polygons;

• NrOfRefPolsWoCandSegms - Number of reference polygons without candidate segments;

• NrOfCandSegments - Number of candidate segments (for which the

• SumUndersegAreaAbsDiff - Sum of absolute undersegmented area differences;

• TotalAreaRefPols - Total area for reference polygons (with one or more candidate segments).

Note

Compared to AssesSeg tool (from Novelli et al., 2017), this function has some slight numerical differences in cases where the number of reference polygons without candidate segments is higher than 0. Probably these derive from area calculations or specific aspects of each implementation.

References

Castilla, G., 2003. Object-oriented analysis of remote sensing images for land cover mapping: conceptual foundations and a segmentation method to derive a baseline partition for classification. PhD Thesis, Polytechnic University of Madrid.

Castilla, G., Hay, G.J., 2008. Image objects and geographic objects. In: Blaschke, T., Lang, S. (Eds.), Object-based Image Analysis: Spatial Concepts for Knowledgedriven Remote Sensing Applications. Springer-Verlag, Berling, Heidelberg, pp. 91-110.

Lang, S., Albrecht, F., Kienberger, S., Tiede, D., 2010. Object validity for operational tasks in a policy context. Journal of Spatial Science 55 (1), 9-22.

Liu, Y., L. Bian, Y. Meng, H. Wang, S. Zhang, Y. Yang, X. Shao, and B. Wang. 2012. Discrepancy measures for selecting optimal combination of parameter values in object-based image analysis. ISPRS Journal of Photogrammetry and Remote Sensing 68:144-156.

Novelli, A., M. Aguilar, F. Aguilar, A. Nemmaoui, and E. Tarantino. 2017. AssesSeg - A Command Line Tool to Quantify Image Segmentation Quality: A Test Carried Out in Southern Spain from Satellite Imagery. Remote Sensing 9:40.

joaofgoncalves/SegOptim documentation built on Feb. 5, 2024, 11:10 p.m.