A hybrid depth estimation method based on light field data

A technology of depth estimation and light field data, applied in image data processing, calculation, image analysis, etc., can solve the problems of high algorithm complexity and accuracy improvement

Active Publication Date: 2019-02-19
AVIC SHANGHAI AERONAUTICAL MEASUREMENT CONTROLLING RES INST
View PDF12 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The parallax-based depth estimation algorithm is based on the principle of binocular parallax for depth estimation. This algorithm simplifies the vid

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
  • A hybrid depth estimation method based on light field data
  • A hybrid depth estimation method based on light field data
  • A hybrid depth estimation method based on light field data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0156] The hybrid depth estimation method based on light field data of the present invention adopts the existing technology to obtain the basic light field camera data, and the light field image can be captured by a Lytro camera, and the focus stack image and the full focus image can be extracted by using LytroDesktop software. Depth of focus is used as a parameter, and the focus stack image within the maximum depth of field range is extracted from the foreground to the background with a step size of 0.2. Use the matlab tool library LFToolbox0.4 to get the sub-image.

[0157] combine figure 1 , a hybrid depth estimation method based on light field data, comprising the following steps:

[0158] Step 1: Acquire light field data: Obtain four-dimensional light field image data through a light field camera, and obtain the focus stack image, sub-image, full-focus image and focus parameters of the scene after processing, where the focus parameters include white image calibration par...

Embodiment approach

[0180] Further, the parallax post-processing, that is, performing denoising processing on the parallax map, the specific implementation method is as follows:

[0181] Using the median filter of m×n window, the formula is as follows:

[0182]

[0183] In the implementation, the disparity value Dn is obtained after m=n=8 median filtering; the disparity map is represented by pixels, and the disparity is converted into a low-resolution depth map Z' by using the camera calibration parameters and the disparity conversion relationship b .

[0184] Step 3: Upsampling based on bilateral filtering: In the process of acquiring sub-images of light field data, the pixel resolution is reduced, and an upsampling process is required; the light field full-focus image has clear texture characteristics, which can improve the quality of the depth map; combined with The full-focus color image of the light field and the small-resolution depth image obtained by parallax estimation are subjected ...

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 a hybrid depth estimation method based on light field data. The method includes: using a hybrid depth estimation method, the focus stack image to perform the hybrid depth estimation based on the parallax and the blur detection on the sub-image and the all-focus image obtained by the light field camera are collected, and the sub-image with the angle difference is used for fusion matching. The method performs parallax analysis, performs focus detection on the simultaneously acquired focus stack image to obtain a depth image, and uses a classification and an adaptive hybrid method to fuse the depth map obtained based on the parallax and the depth map based on the focus; after upsampling the sub image Perform a fusion of depth maps. The invention can obtain the scene depth information by using only one light field camera, and the algorithm is simple and accurate.

Description

technical field [0001] The invention relates to the technical field of stereoscopic image processing, in particular to a hybrid depth estimation method based on light field data. Background technique [0002] As a key technology of 3D reconstruction, depth estimation is a key intermediate step in fields such as stereo display, machine vision, virtual reality, and intelligent vision. Depth measurement or depth estimation used manual measurement in the early days, and later more accurate laser measurement appeared. Now, with the development of image processing technology, image-based depth estimation has become a research hotspot, which has promoted non-contact measurement technology and three-dimensional Development of reconstruction techniques. Image-based depth estimation has the characteristics of simple acquisition equipment and easy integration, and expands the two-dimensional limitation of the original camera image. Depth estimation enables mobile devices to have three...

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): G06T7/55
CPCG06T7/55
Inventor 李贺建黄建民沙龙张镇军刘鹏程冉广奎王贺
Owner AVIC SHANGHAI AERONAUTICAL MEASUREMENT CONTROLLING RES INST
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