Mask image segmentation method and system

Abstract

The invention belongs to the technical field of computer image processing, and discloses a mask image segmentation method and system, comprising the following steps: 1. Mask image detection, extracting the contour, convex hull, convex defect and four elements of the convex defect of the mask image Two, convex defect screening, screening convex defects that meet the following conditions: the distance between the farthest point and the contour is greater than the first threshold, the distance from the starting point to the end point is greater than the second threshold, and the distance between the farthest point and the contour is greater than the same Fifty percent of the maximum distance between the farthest points of all convex defects on the convex hull and the contour; 3. Segmentation processing, find out the distance between the farthest points on the same convex hull in the filtered convex defects The nearest, the farthest point is used as the split point, and the line splitting is sorted along the split points in pairs. A mask image segmentation system is also provided, and the system adopts the aforementioned method to realize the segmentation processing of the mask image. Addresses over- or under-segmentation in mask image segmentation.

Description

technical field [0001] The invention belongs to the technical field of computer image processing, in particular to a mask image segmentation method and system, which can be used for segmentation of cell mask images or other various mask images. Background technique [0002] With the rapid development of imaging technology and image processing technology, people often need to process various images to extract the desired feature information in the image, and various image segmentation algorithms emerge as the times require. [0003] When the elements to be extracted in the image do not overlap and adhere to each other, various image segmentation algorithms can perform image segmentation processing well. However, when the elements to be extracted in the image are dense or small in size, such as when similar objects are clustered together, typical examples are the image of the nucleus mask and the image of some working parts, etc. Since the image obtained by the camera is two-d...

AI Technical Summary

Problems solved by technology

However, when the elements that need to be extracted in the image are dense or small in size, such as the case where similar objects are gathered together, typical examples include cell nucleus mask images and images of some working parts, etc. Since the images obtained by the camera are two-dimensional, Therefore, there are often situations where two or more images of objects are glued together. Such glued images are very unfavorable for the extraction of image feature information, and it is difficult to distinguish them. It is prone to over-segmentation or under-segmentation of the image. In this case, even the simplest count of features in the image will be wrong

[0004] Existing various image segmentation algorithms cannot handle image segmentation with overlapping or cohesive elements well, and over-segmentation or under-segmentation are serious when processing barely, that is to say, these methods have more or less deficiencies. Therefore, it is necessary to improve the image segmentation method to achieve better image segmentation processing effect.

Images