A medium-intelligent segmentation parameter automatic selection method for strengthening a specific area class target attribute

Abstract

The invention relates to a method for automatically selecting medium-intelligent segmentation parameters for strengthening specific area class target attributes. The method comprises the following steps: acquiring image data and selecting a specific area; setting different image segmentation parameter tuples, and calculating segmentation results based on the different parameter tuples; Measuring the intelligent membership degree, uncertainty degree and non-membership degree in the specific region class target based on the rhombic region boundary under each segmentation result; Measuring the intelligent membership degree, uncertainty degree and non-membership degree in the specific region class target based on the square region boundary under each segmentation result; and calculating the middle-intelligence similarity of each segmentation result, and finally determining a segmentation parameter tuple suitable for the current image distribution. The method is simple to implement and widein application range, and based on the current image characteristics, the method can automatically complete the selection of the segmentation parameters for strengthening the attributes of the specific interest region type targets, serves subsequent tasks such as target tracking and the like, and greatly improves the performance of related tasks.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for automatically selecting neutrosophic segmentation parameters that strengthens the attributes of targets in specific regions. Background technique [0002] Image segmentation is a key technology in computer vision analysis. It mainly uses pattern recognition, optimization theory, probability theory, random process, machine learning and other theories to jointly analyze and determine the image segmentation area. Image segmentation is widely used in medical image lesion extraction, object detection, scene understanding, visual object tracking, etc. The segmentation results of existing image segmentation algorithms are usually closely related to parameter initialization, and different initialization parameters correspond to different segmentation results. [0003] Based on the image segmentation results, combined with the class object attributes, the image segme...

AI Technical Summary

Problems solved by technology

Since the accuracy of the class target measurement depends on the segmentation results, in many applications, it is expected to obtain segmentation parameters that strengthen the class target degree of a specific area, but only relying on the maximization of the class target degree of this area is easy to introduce noise, resulting in non-linearity. expected consequences

Images