Image segmentation is an essential step for many computer vision tasks. Evaluating the quality of image segmentations becomes indispensable for choosing an appropriate output of the image segmentation algorithms. To quantitatively evaluate the segmentation quality, various evaluation measures have been proposed to produce a quality map, and a spatial pooling algorithm is followed to combine the quality map into a single quality score. In this paper, we propose two pooling strategies instead of using the conventional spatial average operation. By assigning perceptual meaningful weights to the quality maps, we obtain evaluation measures that are correlated with the human perception of segmentation quality. Specifically, a quality-based and a visual importance-based pooling strategies are designed and tested on some popular evaluation measures, respectively. To the best of our knowledge, this is the first work that applies perceptual pooling strategies for segmentation evaluation. Extensive experiments are conducted on a subjective evaluation benchmark and the Berkeley Segmentation Dataset (BSDS500). The results indicate that the proposed strategies can improve the performance of existing evaluation measures and produce a more perceptually meaningful judgment on the segmentation quality.
Machine Vision and Applications – Springer Journals
Published: Jan 8, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud