Multi-resolution segmentation and object export

Segments GeoTIFF images into homogenous objects using an algorithm from Trimble eCognition.


Description

A processing block that segments an image at a provided scale into image objects and exports them as segmented polygons in a Shapefile.

See this block on the marketplace.

How it works

Use this block as an initial processing step in landcover, vegetation, water, urban area analysis, and change detection workflows.

This block uses an iterative algorithm that merges pairwise objects starting from individual pixels. It merges these objects until the specified homogeneity criteria is reached, and then fits polygons to match these objects.

  • Input: a projected GeoTIFF image with 3–4 spectral bands (RGB+).
  • Output: a Shapefile with 2D+ polygons describing identified image objects with geocoded coordinates.
ParameterOverview
VAR_SPinteger / required
The scale parameter sets the homogeneity criteria. It is a combination of color and shape criteria. Use a higher parameter value to outline larger objects of interest, and vice versa.
There is no input range, and the default value is 42.
VAR_SIMPinteger / required
The simplification parameter sets how the polygons will be drawn around the segmented objects.
The value can range as follows:
  • 0 — drawn polygons will follow pixel outlines of image objects.
  • 1 — drawn polygons will be simplified.
The default value is 1.
VAR_Shapefloat / required
The shape parameter determines the weighting given to shape of objects over their color during segmentation. Use a higher parameter value to increase the influence of shape and lower the influence of colors.
The value can range as follows:
  • A minimum value of 0 — the color of objects will be prioritized.
  • A maximum value of 1 — the shape of objects will be prioritized.
The default value is 0.1.
VAR_Compactnessfloat / required
The compactness parameter determines the weighting given to compactness of objects over their smoothness during segmentation. Use a higher parameter value to increase the compactness and lower the smoothness of objects.
It can range from:
  • A minimum value of 0 — the smoothness of objects will be prioritized.
  • A maximum value of 1 — the compactness of objects will be prioritized.
The default value is 0.5.

To find more about these parameters, see the eCognition website.

Compatibility

This processing block is a powerful tool that can be used as part of an image processing workflow.

Different object types can be segmented by manipulating the parameters. The objects are exported as Shapefiles. Shapefiles are a geospatial vector data format that can be imported to most GIS software.

The objects can then be reshaped, classified and analyzed.

Example

An example is shown below that identifies image objects using Pléiades catalog data from storage as a data source.

{
  "up42-processing-from-storage:1": {
    "asset_ids": ["d86b5212-e3c3-4807-8715-324171a79833"]
  },
  "ecognition_mrs:1": {
    "VAR_SP": "100",
    "VAR_SIMP": "1",
    "VAR_Shape": "0.1",
    "VAR_Compactness": "0.5"
  }
}
Sample GeoTIFF input
Sample GeoTIFF input
Sample Shapefile output
Sample Shapefile output

Capabilities

Input

raster
up42_standard
bands
{
  "or": [
    [
      "red",
      "green",
      "blue"
    ],
    [
      "red",
      "green",
      "blue",
      "nir"
    ],
    [
      "red",
      "green",
      "blue",
      "nir",
      "pan"
    ]
  ]
}
dtype
{
  "or": [
    "uint16",
    "uint8"
  ]
}
formatGTiff
sensor
{
  "or": [
    "Pleiades",
    "SPOT",
    "Sentinel2",
    "Landsat8",
    "Hexagon"
  ]
}
resolution
{
  "or": [
    0.15,
    0.3,
    0.5,
    1,
    1.5,
    10,
    15
  ]
}
processing_level
{
  "or": [
    "l2",
    "l3"
  ]
}

Output

To know more please check the block capabilities specifications.