An Objective Metric for the Evaluation of Image Segmentation Quality Using Human Visual System Characteristics
An Objective Metric for the Evaluation of Image Segmentation Quality Using Human Visual System Characteristics, Proc International Workshop on Video Processing and Quality Metrics for Consumer Electronics - VPQM, Scottsdale, United States, Vol. ., pp. . - ., January, 2009.
Digital Object Identifier:
The assessment of segmentation quality results assumes a crucial importance to evaluate to which degree the targets of a segmentation algorithm are met.
The current practice for segmentation quality evaluation mainly consists in subjective ad hoc assessment by a representative group of human viewers, a time-consuming and expensive process. The availability of an automatic procedure, objective segmentation quality evaluation, able to mimic the results of a formal subjective evaluation, would be preferred. Such an objective evaluation algorithm could then be used to compare the performance of different segmentation algorithms, or several configurations of the same algorithm, in view of the requirements characterizing a given application scenario.
This paper contributes to the field of image segmentation quality evaluation, proposing a new perceptual objective metric to assess the quality of segmented objects, by comparison to a reference segmentation, i.e., a relative evaluation metric.