Filters a label map by rules (e.g., remove large or deformed objects).

The properties of the labeled image regions (features) must be provided in a specific tabular format (columns: label, feature1, feature2, and so on, where feature1 and feature2 can be arbitrary strings). Each row corresponds to a labeled image region. An example is given below.

label	area	eccentricity
1	344	0.42521053699241596
2	434	0.47679001553231926
3	907	0.9973539531125177
4	14320	0.17131009631035327

The rules also must be supplied in a specific tabular format with three rows. The top-left cell is empty, and the rest of the first row corresponds to the feature names (such as feature1 or feature2, see above). The rest of the first column corresponds to the two values min and max. Each of the rows defines the minimum and maximum values for the corresponding features. A labeled image region is retrained if and only if it passes all checks with repsect to the given min and max values. An example is given below.

	area	eccentricity
min	500
max	100000	0.5