A method and device for image segmentation

Abstract

The invention provides an image segmentation method and device. The method includes: acquiring the same first and second images to be processed; preprocessing the first image to be processed and converting it into a gradient image; acquiring the gradient image on each first area. The set of marker points and the coordinate values ​​and grayscale values ​​of the pixels of the gradient image; based on the above parameters, the pre-constructed foreground grayscale histogram and background grayscale histogram, determine each pixel on the contour line of the target area to be segmented The coordinate value of the point, and based on this, on the second image to be processed, the color of the area other than the target area to be segmented is set to a color different from the target area to be segmented, so as to achieve segmentation of the target area to be segmented. . The invention can achieve the purpose of rapid and accurate segmentation.

Description

technical field [0001] The invention relates to the technical field of detection, in particular to an image segmentation method and device. Background technique [0002] With the development of my country's economy and society, people's quality of life has been significantly improved. The demand for meat products is not only satisfied in quantity, but also has high requirements for the quality of meat products. The quality and safety of pork products are closely related to the people's expectations The quality of life is closely related. Pig breeding is not only related to the development of my country's economy, but also related to people's livelihood and national economy and social stability. However, epidemics pose a serious threat to pig breeding and production. Therefore, segmenting specific pig images is of great significance for future pig target detection, recognition, and disease diagnosis and treatment. With the development of image processing technology, image seg...

AI Technical Summary

Problems solved by technology

The watershed algorithm has good robustness, but it is easily affected by noise; the over-segmentation phenomenon of the adaptive watershed algorithm is very serious, which can easily lead to the loss of image target information or too much segmentation; the LazySnapping algorithm performs segmentation, which can realize the extraction of specific colors

[0004] For segmentation objects with single color, fuzzy edges and low contrast, complex background objects may be similar to target feature values, and the problem that the specific target that the user wants to select is not easy to segment is still unsolved.

Images