Image segmentation method and apparatus

Abstract

The invention discloses an image segmentation method and a device. The method comprises acquiring an image to be segmented and a background image unver a first shooting angle of view, the first shooting angle of view being one of a plurality of shooting angles of view; acquiring an image to be segmented and a background image under the remaining shooting angles of the plurality of shooting angles;obtaining a visual shell according to the obtained images to be segmented and background images under all shooting angles; acquiring an actual silhouette contour of the target object according to theprojection of the visual housing in an image coordinate system corresponding to the first photographing angle of view; an image of the target object at the first shooting angle being obtained from the actual silhouette and the image to be segmented at the first shooting angle. The device comprises a first shooting angle image acquiring module, a remaining shooting angle image acquiring module, avisual housing generating module, an actual silhouette outline acquiring module and a target object image acquiring module, wherein the first shooting angle image acquiring module comprises a first shooting angle image acquiring module, a remaining shooting angle image acquiring module, a visual housing generating module, an actual silhouette outline acquiring module and a target object image acquiring module. The invention improves the accuracy of image segmentation through the technical scheme.

Description

technical field [0001] The invention belongs to the technical field of computer vision, in particular to an image segmentation method and device. Background technique [0002] With the rapid development of deep learning in the field of computer vision, more and more computer vision problems rely on deep learning to obtain more mature and reliable solutions. Due to the large solution space, deep learning often requires a large amount of data to train parameters, and the cost of manual data collection and labeling is high. Therefore, it is necessary to propose an automatic image segmentation method to assist data collection and labeling. Existing image segmentation methods have low accuracy. Contents of the invention [0003] In order to at least solve the problem of low image segmentation accuracy in the prior art, the present invention provides an image segmentation method on the one hand, which includes: Step S1, acquiring the image to be segmented and the background ima...

AI Technical Summary

Problems solved by technology

[0003] In order to at least solve the problem of low image segmentation accuracy in the prior art, the present invention provides an image segmentation method on the one hand, which includes: Step S1, acquiring the image to be segmented and the background image at the first shooting angle of view, so The first shooting angle of view is one of the plurality of shooting angles of view; step S2, obtain the image to be segmented and the background image under the remaining shooting angles of the plurality of shooting angles of view; step S3, according to all acquired shooting angles of view Obtain a visible shell from the image to be segmented and the background image; step S4, obtain the actual silhouette of the target object according to the projection of the visible shell in the image coordinate system corresponding to the first shooting angle of view; step S5, according to the The actual silhouette and the image to be segmented under the first shooting angle of view are used to obtain the image of the target object under the first shooting angle of view; wherein, the material of the bearing plate carrying the target object is a transparent material, and a plurality of the shooting angles The viewpoints where the viewing angles are located include the viewpoints above and below the carrier board

Images