The main segmentation algorithms are analyzed in the paper: 
threshold segmentation, morphological segmentation, segmentation 
of region growing, cluster segmentation. 
Objective and subjective quality criteria for segmentation are examined.
To quantify the quality of segmentation it is analyzed two criteria - 
supervisory and nonsupervisory. The article focuses 
on the use supervisory criteria based on 
calculating distances between images using metrics. 
Image is generally represented as a union of contours and regions. 
For comparison, contours of images used Frechet metric. 
In this paper, the algorithm of determining the upper limit of Frechet distance, 
which is based on the use of contour characteristic points: 
points maximum, minimum, and maximum curvature and inflection points. 
To compare regions of images used Gromov - Hausdorff metric 
that considers isometric transformation of images. 
For comparison, most images used a combination of these two metrics. 
As standard images used images that expert segmented. 
In carrying out computer experiments quantitative quality 
of segmentation algorithms used segmentation base histological 
images of breast cancer capacity of about 2000 images.