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

Method for matching depth image

A depth image matching and depth technology, applied in the field of depth image matching, can solve problems such as easy to fall into local optimum, and achieve the effect of easy recognition, little influence by noise, and simple calculation

Active Publication Date: 2009-03-11
SHENZHEN ESUN DISPLAY
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The disadvantage of this algorithm is that it needs to give initial position estimates of two depth images, and it is easy to fall into local optimum
This method only uses the Harris corner points of the texture image to find the feature points, and the matching relationship is completed through geometric features, and does not use texture information to find the matching relationship.

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
  • Method for matching depth image
  • Method for matching depth image
  • Method for matching depth image

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0063] The object to be matched in this specific embodiment is a regular-shaped vase, such as Figure 2a , 2b , 3a, and 3b, the first texture map T of the vase was collected from two fields of view with an included angle of 36° 1 , the second texture map T 2 and the first depth like P 1 , the second depth image P 2 . And get the first texture map T 1 with the first depth like P 1 Correspondence between and the second texture map T 2 with a second depth like P 2 corresponding relationship.

[0064] Such as Figure 4 As shown, in the first texture map T 1 Select the first texture map T from 1 and the second texture map T 2 The rectangular overlapping region R of . Such as Figure 5 As shown, a Gaussian filter is performed on the rectangular overlapping area R, and the Sobel operator is applied for edge extraction.

[0065] Such as Figure 6a As shown, the rectangular overlapping region R is divided into a plurality of uniform units. Each unit is scanned line by li...

specific Embodiment approach 2

[0086] The difference between this specific embodiment and specific embodiment 1 is:

[0087] The object to be matched in this specific embodiment is a small plastic figure with a free-form surface, and its depth data and texture data are collected from two fields of view with an included angle of 30°, and the matching of depth images is completed by using the method of the present invention. Figures 8a and 8b are texture images of two fields of view respectively, and Figs. 9a and 9b are depth images of two fields of view respectively. As shown in Figures 10a and 10b, this embodiment collects 10 pairs of texture feature points. The correlation coefficient and Hausdorff distance of these texture feature point pairs are shown in Table 2:

[0088] Table 2

[0089]

[0090] The Hausdorff distances of the 5th, 8th, and 10th feature point pairs in Table 2 are greater than the threshold h (13.34 pixels), and they are considered invalid. Select 5 values ​​with ...

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

A method for matching depth image includes selecting depth image and texture image of two fields on 3-D object to obtain corresponding relation depth image to texture image at two fields, selecting texture character point pair set from texture image of two fields, obtaining depth point pair set corresponding to said texture character point pair set according said corresponding relation and said character point pair set and calculating out rotary matrix and translation matrix according to obtained depth point pair set.

Description

technical field [0001] The invention relates to three-dimensional digital imaging and modeling, in particular to a depth image matching method. Background technique [0002] The 3D digital imaging and modeling of entities has important research and application value in computer vision, pattern recognition, 3D animation and other fields. Generally speaking, realistic 3D modeling includes the following four main steps: (1) Acquisition of depth images and texture maps under different fields of view. Among them, the texture map is commonly referred to as an optical image, and the value of each pixel represents the intensity of reflected light at each point on the surface of the object. The depth image (Range Image) is different from the optical image. The value of each pixel in the depth image does not represent the intensity of reflected light at each point on the surface of the object, but the distance between each point of the object and the sensor. Usually, a 3D measuremen...

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 Patents(China)
IPC IPC(8): G06T7/00
Inventor 彭翔刘晓利田劲东
Owner SHENZHEN ESUN DISPLAY