Accuracy Assessment Measures for Object-based Image Segmentation Goodness

By N. Clinton, A. Holt, J. Scarborough, Li Yan, Peng Gong. 2010

347 citations recorded by the source corpus.

To select an image segmentation from sets of segmentation results, measures for ranking the segmentations relative to a set of reference objects are needed. We review selected vector-based measures designed to compare the results of object-based image segmentation with sets of training objects extracted from the image of interest. We describe and compare area-based and location-based measures that measure the shape similarity between segments and training objects. By implementing the measures in two object-based image processing software packages, we illustrate their use in terms of automatically identifying parsimonious parameter combinations from arbitrarily large sets of segmentation results. The results show that the measures have divergent performance in terms of the identification of parameter combinations, Clustering of the results in measure space narrows the search. We illustrate combination schemes for the measures for generating rankings of segmentation results. The ranked segmentation results are illustrated and described.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/abe7a5b72f5d5e75042c4f69f15f8f17f4b11426

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Peng Gong
2. Congcong Li
3. Jie Wang
4. Lei Wang
5. Luanyun Hu

In Media Technology

1. Peng Gong
2. Darius Phiri
3. George Alan Blackburn

Citations in the corpus

• Comparison of Classification Algorithms and Training Sample Sizes in Urban Land Classification with Landsat Thematic Mapper Imagery (2014, background): “For the segmentation, the threshold is the most important parameter, which determines the size of the objects [38].”
• Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN (2023, background): “Where predictions were not accurate, it was slightly more likely to be from under-segmention (0.23 – 0.45) than over-segmentation (0.13 – 0.23) (Clinton et al., 2010) across all sites (see “Extended results”) (Fig.”
• A Survey on Object Detection in Optical Remote Sensing Images (2016, neutral): “…number of researchers have proposed several object-based metrics for assessing both geometric and thematic accuracies of object-based image classification (Albrecht, 2010; Clinton et al., 2010; Lizarazo, 2014; MacLean and Congalton, 2012; Radoux et al., 2011; Zhan et al., 2005; Zhen et al., 2013).”
• Unsupervised image segmentation evaluation and refinement using a multi-scale approach (2011, neutral): “Dissimilarity measures can be based on differences in color, location, size, or shape (Abeyta and Franklin, 1998; Carleer et al., 2005; Moller et al., 2007; Neubert et al., 2008; Clinton et al., 2010).”
• Geospatial Resolution of Human and Bacterial Diversity with City-Scale Metagenomics (2015, neutral): “…a grouping in reported ethnicity (self-reported as White, Black, Asian, Hispanic) from all areas of an imagesegmented U.S. Census Map (Figure S4) (Clinton et al., 2010), then compared these to samples wherein we observed enough human-mapping reads to call variants (see Supplemental Experimental…”
• Efficient corn and soybean mapping with temporal extendability: A multi-year experiment using Landsat imagery (2014, neutral): “The “D metric” provided by the software indicates the degree of discrepancy between segmented images and a reference polygon layer whichwas manually created to represent actual field boundaries (Clinton et al., 2010).”