Circular masking-out area rate determination-based adhesive particle image concave point segmentation method

Abstract

The present invention relates to a circular masking-out area rate determination-based adhesive particle image concave point segmentation method. The method comprises the following steps of 1) carrying out the image pre-processing to obtain a binary image of a particle image; 2) carrying out the concave point rough detection to obtain an angular point image of the particle image; 3) carrying out the concave point accurate detection, and utilizing an area rate method to obtain all concave points capable of being used for the particle segmentation in a regional contour; 4) carrying out the concave point pairing, wherein the selected concave point pairs are used as the segmentation points of an adhesive particle image; 5) constructing a segmentation line of the particles, obtaining the contour coordinates of individual particles, and combining the coordinates of two segmentation points to obtain the complete particle contour. According to the present invention, by combining an angular point detection method and a method based on concave point analysis, an operand problem brought by purely utilizing the concave point search based on area is avoided; by setting few parameters, a lot of sample training is not needed; at the same time, the segmentation paths can be replanned, thereby being able to adapt to the different shape and size change of the images.

Description

technical field [0001] The invention relates to the technical field of cohesive particle image segmentation, in particular to a cohesive particle image concave point segmentation method based on circular mask area ratio discrimination. Background technique [0002] Image-based particle analysis is an important technology in the field of image processing, which is widely used in cell analysis, chromosome segmentation, pathogen detection, seed parameter analysis, nanoparticle segmentation, rock particle analysis and other fields. Generally speaking, the target object and the background in the captured image have a high contrast, and the method of extracting the target object from the background can be realized through threshold segmentation. However, in real scenes, due to the contact, fusion or occlusion between particles, the particles in the image are cohesive to each other, and the contour information of individual particles is hidden, which affects the accuracy of paramet...

AI Technical Summary

Problems solved by technology

This method only considers the angle of the concave point and does not consider the depth, which is susceptible to interference

Although this method can achieve good results in a certain field, due to the complexity of the problem of adhesion or occlusion between particles, the analysis method based on pits still has the defects that pit positioning is easily disturbed and pit pairing is difficult. The structure of the separation line does not match the actual shape of the object, and cannot reflect the initial state of the particle

Images