Final Report for the SAREO Project: Evaluation Methods and Analysis of Segmentation Algorithms for SAR and EO Imagery

PDF

Authors
  1. Markle, B.
Corporate Authors
Defence R&D Canada - Ottawa, Ottawa ONT (CAN);Array Systems Computing Inc, Toronto Ont (CAN)
Abstract
In this project we are primarily concerned with developing a strategy by which to assess and evaluate the performance of segmentation algorithms on SAR&EO images. In the current document we present baselines for both Scene and Target segmentation evaluation. The baselines include Relative and Standalone evaluation for both Scene and Target segmentation evaluation: - Scene segmentation evaluation in the absence of ground truth – by using the so-called “Standalone” evaluation methods; - Scene segmentation evaluation in the presence of ground truth – by using the so-called “Relative” evaluation methods; - Target segmentation evaluation, in the presence of ground truth – by using the so-called “Relative” evaluation methods; - Target segmentation evaluation, in the absence of ground truth – by using the so-called “Standalone” evaluation methods. In our opinion, these baselines are suitable for evaluating both SAR and EO imagery. This document also demonstrates usage of the SAREO Matlab toolbox developed under this contract, and includes installation instructions for the code.
Report Number
DRDC-OTTAWA-CR-2006-007;ASC-SAREO-005 — Contractor Report
Date of publication
01 Jan 2006
Number of Pages
138
DSTKIM No
CA027085
CANDIS No
525119
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: