Intelligent measurement method based on AR and multiple semantic segmentation

Abstract

The invention discloses an intelligent measurement method based on AR and multiple semantic segmentation, and the method comprises the steps: obtaining the plane mapping G(x, y, z) of a to-be-measured region image through employing an AR modeling method; performing a first round of semantic segmentation on the to-be-measured area image after the anomaly detection processing to obtain a local image of the to-be-measured object; performing second-round semantic segmentation on the local image of the to-be-measured object obtained in the step S3 to obtain a semantic entity of a region where a measurement point is located, and performing normalization processing on the obtained semantic entity to obtain a two-dimensional coordinate P(x, y) of the to-be-measured point; combining the plane mapping G(x, y, z) of the to-be-measured area image obtained in the step S1, and obtaining the size parameter of the to-be-measured object according to the two-dimensional coordinate P(x, y) of the to-be-measured point. According to the method, the measurement points of the object can be obtained by adopting a multi-semantic segmentation method, the interference of the surrounding environment is eliminated, the automatic acquisition of the measurement points is automatically realized, and the measurement points obtained by multiple times of segmentation are more accurate than manual selection.

Description

technical field [0001] The invention relates to the technical fields of augmented reality and computer vision, in particular to an intelligent measurement method based on AR and multiple semantic segmentation. Background technique [0002] In production and life, it is often necessary to measure the length and area of ​​real objects. There are two main measurement methods: manual measurement and instrument measurement. Manual measurement mainly uses tape measure and ruler for manual measurement, which is time-consuming, laborious and inefficient; instrument measurement is carried out by laser scanner, ultrasonic technology, etc. These methods have high precision, but the equipment is expensive, which is not conducive to large-scale application. At present, there are also methods that use computer vision for measurement. This type of method has the advantages of simple operation and labor-saving. [0003] Existing object measurement methods based on computer vision usually ...

AI Technical Summary

Problems solved by technology

Since the screen size is generally not very large, manual selection of points to be measured is easily affected by the surrounding environment, and each person's clicking habits are different, which will exacerbate the inaccuracy of the measurement results

Images