This is an automated version of the Estimation of Scale Parameter tool (ESP). The tool works on multiple layers (maximum of 30) and automatically produces three scale levels, based on the concept of Local Variance (Woodcock and Strahler, 1987).

ESP_Tool.PNG

Status of work: Public Domain

The download package contains:

  • ESP2_Estimation_Scale_Parameter_2.dcp (encrypted eCognition rule set)
  • ESP2_User_Guide.pdf
  • ESP_Estimation_Scale_Parameter_Chart.exe (a stand-alone tool for visualizing and interpreting the results. This tool is programmed in .NET, therefore the .NET framework needs to be installed on your machine if you want to interpret the results (optional, for advanced users).
  • ZedGraph.dll (a dynamic link library which is needed to run the ESP_Estimation_Scale_Parameter_Chart.exe tool)

Reference:  Drăguţ, L.; Csillik, O.; Eisank, C.; Tiede, D. Automated parameterisation for multi-scale image segmentation on multiple layers. ISPRS J. Photogramm. Remote Sens. 2014, 88, 119–127, doi:10.1016/j.isprsjprs.2013.11.018. https://www.sciencedirect.com/science/article/pii/S0924271613002803

Download 'ESP II' Tool 

5 comments

  • Christian Weise

    ┗(ツ)┓┗(ツ)┛ Thanks L. Drăguţ & O. Csillik & C. Eisank & D. Tiede! ┏(ツ)┛┏(ツ)┓

  • Stephanie Stephanie

    hi . how come it says "You do not have read access to the process?" I was trying out this algorithm but I encountered this error. Please help

  • Keith Peterson

    This is a customized algorithm and it has been encrypted by the developer. You can execute it but you cannot open it to see the detailed content/source code. Do you have a concrete question about how the algo. works?

  • Stephanie Stephanie

    I tried using this tool recently, following the tutorial I have seen on youtube. However, I was surprised that the processing took 12 hours to finish. This is way beyond the example they did on youtube which took like 10 mins to finish. Will greatly appreciate it if anyone could address this.

  • Keith Peterson

    That is because this is designed as an exploritory tool, not to be run on your entire data set, rather a small subset. It helps to determine what scale parameter is best suited to your data. Therefore, run it on a small subset, determine the ideal scale parameter (according to this algo) and then plug that into your MRS algo and run that on your entire full dataset. Depending on the size of your full dataset, you may want to consider a tile scheme in combination with eCognition Server for better performance. Or try distributing on more cores, but only if you have the RAM to back it up.

Your Answer

Articles in this section

Need more help?

Enter Community Contact Support