Unlock instant, AI-driven research and patent intelligence for your innovation.

Phase analysis method based on multi-scale attention mechanism network

An analysis method and attention technology, applied in neural learning methods, biological neural network models, image analysis, etc., can solve the problems of lack of phase unwrapping methods and few reports of weak textures

Active Publication Date: 2021-09-10
SICHUAN UNIV
View PDF7 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In addition, these methods are mostly based on simulation data or biomedical data for continuous phase prediction, but there are few reports on deep learning phase analysis of real faces with weak textures.
Therefore, there is still a lack of a phase unwrapping method with both measurement accuracy and measurement efficiency for the situation where the surface shape of the measured target is relatively complex and the phase information changes drastically, especially in the case of spatial isolation, noise, shadow and undersampling areas.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Phase analysis method based on multi-scale attention mechanism network
  • Phase analysis method based on multi-scale attention mechanism network
  • Phase analysis method based on multi-scale attention mechanism network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] figure 1 A schematic flow diagram of a phase resolution method based on a multi-scale attention mechanism network is shown, and the method includes the following steps:

[0050] S1, projecting a triple-frequency N-step phase-shifted fringe image to the object to be measured, and the camera synchronously captures the deformed fringe image modulated by the surface shape of the object to be measured.

[0051] The light intensity map of any group of N-step (N≥3) phase-shifted grating images of the projector can be expressed as:

[0052]

[0053] In formula (1), i=1,...,N represents the i-th step phase shift (N≥3). I i (x,y) represents the light intensity function on the time axis, a(x,y) represents the background light intensity, and b(x,y) represents the modulation depth of the stripes. f 0 is the carrier frequency, Indicates the modulated phase of the target point caused by the height distribution in the i-th fringe image.

[0054] S2. Convert the deformed fring...

Embodiment 2

[0083] In order to verify the phase unwrapping method provided by this embodiment of the present invention, a specific experimental case of the method is exemplarily given below.

[0084] figure 2 A schematic diagram of an experimental case of a complex surface reconstruction system based on binocular vision is shown according to the invention, and the experimental conditions are set as follows:

[0085] (1) 2 completely isolated measured targets: including 1 mask and 1 face model. (2) Projector: frame rate 120fps, resolution 1280*1024, every 12 structured light images as a group, continuous loop projection, synchronously provide external trigger signal to camera; (3) Camera: including left camera 715 and right camera Camera 716, with a frame rate of 120fps and a resolution of 1280*1024, receives the external trigger signal of the projector and shoots two measured targets illuminated by structured light. There is a fixed baseline distance between the projector and the left a...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to the field of optical three-dimensional measurement, in particular to a phase analysis method based on a multi-scale attention mechanism network, which comprises the following steps of: S1, projecting a three-frequency N-step phase shift fringe image to a measured object, and synchronously shooting a deformed fringe image modulated by the surface shape of the measured object by a camera; S2, converting the deformed fringe image into truncation phase diagrams with different frequencies based on a phase shift method; S3, extracting the truncated phase diagram with the highest frequency from the truncated phase diagrams with different frequencies, inputting the truncated phase diagram with the highest frequency into a pre-trained multi-scale attention mechanism network, and outputting a corresponding continuous phase diagram; and S4, mapping the continuous phase diagram into a three-dimensional surface shape depth to obtain three-dimensional point cloud data of the measured object. By adopting the method, the continuous phase can be quickly predicted by inputting the truncated phase image into the pre-trained network model, error diffusion and accumulation can be inhibited, and the method is not influenced by regions such as discontinuous actual phase and glasses wearing.

Description

technical field [0001] The present invention relates to the field of optical three-dimensional measurement, in particular to an efficient phase analysis method based on a multi-scale attention mechanism network. Background technique [0002] In recent years, 3D reconstruction has been a hot topic in computer vision. Many studies have constructed 3D models by simulating the human binocular vision system, using the parallax generated by observing the same object from the left and right perspectives combined with the principle of triangulation to obtain depth information, which is widely used in machines. In the fields of intelligence, automatic driving, industrial inspection, reverse engineering, virtual reality, 3D manufacturing and 3D printing. Although passive stereo matching has achieved good results, its 3D reconstruction effect is not ideal for objects containing large areas of weak texture areas and scenes with high modeling accuracy requirements. Therefore, it is nece...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T7/521G06N3/04G06N3/08
CPCG06T7/521G06N3/08G06T2207/10028G06T2207/20081G06T2207/20084G06T2207/30201G06N3/045
Inventor 段智涓朱江平黄怡洁游迪
Owner SICHUAN UNIV