This paper reviews the research progress on image segmentation evaluation

especially on evaluation methods and evaluation criteria. Most evaluation methods proposed so far can be categorized into direct methods and indirect methods. The direct methods study segmentation algorithms themselves while indirect methods evaluate the performance of algorithms according to the quality of segmented images. Both direct methods and indirect methods need certain evaluation criteria for performing evaluation tasks. This paper makes a classification of different existing criteria and provides an analytical comparison about these criteria. In addition

for most commonly used evaluation criteria

an experimental comparison has been carried out. A rank of these criteria has been obtained through the analysis of their performance and behavior.This study provides a useful basis for correctly applying different evaluation methods in practical applications and indicates several further research directions.