A survey on evaluation methods for image segmentation

TLDR: This study is helpful for an appropriate use of existing evaluation methods and for improving their performance as well as for systematically designing new evalution methods.

About: This article is published in Pattern Recognition.The article was published on 1996-08-01 and is currently open access. It has received 1117 citations till now.

Figures

Table 1: Comparison results of different evaluation methods.

Frequently asked questions

Q1. What are the properties of segmentation algorithms that can be obtained by analysis?

Other properties of segmentation algorithms that can be obtained by analysis include the processing strategy, processing complexity and efficiency and segmentation resolution of algorithm.[17-18] 

Q2. What are the empirical methods used to evaluate the performance of algorithms?

The empirical discrepancy methods compare the segmented image or output image to the reference image and use their difference to assess the performance of algorithms. 

Q3. What is the way to compute the amount of busyness for a thresholded image?

In practice, they compute the amount of busyness for a thresholded image by using the gray-level co-occurrence matrix of the image.[22] 

Q4. What is the main reason why the blurring effect is not used in the evaluation of algorithms?

In addition, the noisy effect, a very important and common degradation factor influencing the performance of algorithms, cannot be studied by such a method.