Hierarchical processing method of image scene based on Bayesian inference

Abstract

The invention discloses a hierarchical processing method of an image scene based on Bayesian inference. The method comprises the following steps of: defining five shielding indexes including semantics, location, outline, public boundary and interface point based on the image scene semantics marking result, and learning the probability distribution of the shielding indexes; establishing a Bayesianinference framework, and defining a weighted directed graph based on prior probability and likelihood probability; and finally, solving a hierarchical structure of the image scene on the weighted directed graph by use of a hierarchical sorting inference algorithm provided by the invention in combination with the posterior probability maximization. The method disclosed by the invention can be usedfor shielding the image or video virtual scene, generating an image or video virtual scene and the like.

Description

technical field [0001] The invention belongs to the fields of image processing and computer vision, in particular to a layered image scene processing method based on Bayesian reasoning. Background technique [0002] The virtual scene generation technology based on video material is an important part of virtual reality, and it is also a research hotspot in the organic intersection of virtual reality, augmented reality, computer vision and related research directions. Image virtual scene generation is an important part of video virtual scene generation, and the hierarchical occlusion processing of image virtual scene is the key problem that image virtual scene generation technology needs to solve. In the process of virtual scene generation, it is necessary to deal with various spatial occlusion relationships between scene objects, scene objects and video scenes, so as to ensure the correct spatial position relationship of the video virtual scene. Relative hierarchical positio...

AI Technical Summary

Problems solved by technology

These methods study the construction of the hierarchical structure at the pixel level, but only use the information of the middle and low layers in the image to solve the problem of image scene layering, and do not infer the layering results that are most likely to represent its hierarchical structure in the image scene.

Images