Scene semantic segmentation system and method based on light field

Abstract

The invention relates to a scene semantic segmentation system and method based on a light field. The system is composed of a light field image preprocessing module, an image feature extraction module, a scene semantic segmentation module, a data storage module, a display module and a system management module. According to the invention, high-accuracy scene semantic segmentation is realized by using the advantages of the light field image in scene three-dimensional information extraction.

Description

technical field [0001] The invention belongs to the field of image technology, and in particular relates to a light field-based scene semantic segmentation system and method. Background technique [0002] Scene semantic segmentation technology is widely used in autonomous driving, human-computer interaction, image search and other fields. Accurate scene semantic segmentation method is of great significance to scene understanding and its application. Current semantic segmentation methods can be broadly classified into two categories. One is to rely only on a single image for semantic segmentation. For example, Zhang et al. proposed a single-image semantic segmentation method based on the context coding module [1], which uses a dilated convolution strategy to achieve better semantic segmentation on images with simple structures. effect, but these single-image-based semantic segmentation algorithms are generally difficult to achieve good semantic segmentation results in comple...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the problem of insufficient accuracy of existing scene semantic segmentation systems, the present invention provides a light field-based scene semantic segmentation system and method, which fully utilizes the advantages of light field images in scene three-dimensional information extraction to achieve high-accuracy scenes semantic segmentation

Images