Asymptotic global matching binocular parallax acquisition method and system

A binocular disparity and acquisition method technology, applied in the field of computer perspective, can solve problems such as large computing resources and memory resources, finding the optimal matching disparity value, and local matching errors, so as to improve computing efficiency, reduce computing load, The effect of reduced calculation

Pending Publication Date: 2021-02-09
ZHEJIANG SCI-TECH UNIV
View PDF6 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The global aggregation cost matching accuracy is higher, but consumes huge computing resources and memory resources, and the frame rate that can be achieved is very limited; while the local matching speed is faster, but its accuracy is low, and local matching errors will occur, or local missing Optimizing the matching disparity value to produce a large number of discretely distributed mis-match regions in areas where matching features are not obvious

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
  • Asymptotic global matching binocular parallax acquisition method and system
  • Asymptotic global matching binocular parallax acquisition method and system
  • Asymptotic global matching binocular parallax acquisition method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0046] This embodiment provides a specific implementation method for optimizing parallax acquisition.

[0047] Such as image 3 As shown in step 105, the method for global matching of the left image and the right image includes:

[0048] Step 301: Set the search range for the right image with reference to the first pixel-level initial disparity map. For example, the search range may be a parallax range of ±10.

[0049] Step 302: By means of global matching, search for an optimal match with the point on the left image in the search range, and obtain the relative disparity. Specifically, the point in the left image is used as the target point, and the search range of the right image is determined through the first equal-pixel initial disparity map, and the optimal matching point is searched in the right image.

[0050] Step 303: Obtain the optimal disparity according to the relative disparity and the initial disparity in the first iso-pixel initial disparity map. The optimiz...

Embodiment 2

[0070] This embodiment provides a method for multi-level asymptotic matching processing.

[0071] Take the second-level asymptotic matching processing as an example, such as Figure 6 As shown, the method of obtaining binocular disparity through the second initial disparity:

[0072] Step 601: Sampling the rows and columns of the left image at intervals to obtain a second left image with low resolution, and the resolution of the second left image is higher than that of the first left image.

[0073] Step 602: Sampling the rows and columns of the right image at intervals to obtain a second right image with low resolution, the resolution of the second right image is higher than that of the first right image.

[0074] Step 603: The first initial disparity map is interpolated to obtain a second equal-pixel initial disparity map having the same pixel array as the second left image or the second right image. The second equal-pixel initial disparity map may have the same pixel arra...

Embodiment 3

[0081] Such as Figure 7 As shown, this embodiment provides a method for preprocessing the left image or the right image:

[0082] Step 701: Align the left image and the right image in the Y direction.

[0083] Step 702: Crop the left image and the right image according to matching requirements.

[0084] When the binocular image is collected, although the position of the binocular camera is adjusted and centered, the taken pictures are aligned on the Y axis, and there is parallax in the X direction. Usually, there will be operational errors, so that the left image and the right image Not aligned on the Y axis. By clipping the information beyond the matching range, it can ensure the basic conditions of stereo matching and reduce the amount of computation.

[0085] In a specific embodiment, when calculating the BT cost, first calculate the parallax range of the right image, and then calculate the left image. The parallax range of the right image in the X direction must be gre...

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 discloses an asymptotic global matching binocular parallax acquisition method and system, and the method comprises the steps of carrying out the interval sampling of rows and columns ofa left image, and obtaining a first left image with low resolution; performing interval sampling on rows and columns of the right image to obtain a first right image with low resolution; matching thefirst left image and the first right image based on a global matching method to obtain a first initial disparity map; performing interpolation processing on the first initial disparity map to obtain afirst equal-pixel initial disparity map having the same pixel array as the left image or the right image; and taking the first equal-pixel initial disparity map as a reference, and performing globalmatching on the left map and the right map in a set relative disparity range to obtain an optimized disparity map. The first equal-pixel initial disparity map is used as a reference, so that the search range of global matching is conveniently reduced, the calculation efficiency is improved, calculation resources and memory resources are saved, and a high-precision optimized disparity map is obtained.

Description

technical field [0001] The invention relates to the technical field of computer perspective, in particular to a binocular disparity acquisition method and system for asymptotic global matching. Background technique [0002] Binocular vision matching is a machine technology that recovers spatial depth information by computing from a pair of pictures taken from different angles. This technology has received widespread attention in recent years and has become one of the foundations of many intelligent technologies such as autonomous driving, robot vision, virtual / augmented reality, and industrial automation measurement. Algorithms that are widely used at present include: interactive pixel difference (BT) cost algorithm and quasi-global aggregation cost matching (SGM, Semiglobal Matching) algorithm. The BT cost algorithm is a commonly used local algorithm (S. Birchfield and C.Tomasi, "Depth Discontinuities by Pixel-toPixel Stereo," Proc. Sixth IEEE Int'l Conf. Computer Vision, ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/33G06T5/00G06T5/50G06T3/40
CPCG06T7/33G06T5/002G06T3/4007G06T5/50G06T2207/10004G06T2207/20216G06T2207/20228
Inventor 袁嫣红袁海骏罗宏利
Owner ZHEJIANG SCI-TECH UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap