CN105869157A - Multi-lens stereoscopic vision parallax calculating method - Google Patents

Abstract

The invention discloses a multi-lens stereoscopic vision parallax calculating method, and belongs to the field of computer vision. The method comprises four steps: initialization, the obtaining of all binocular parallaxes for range finding, the obtaining of a multi-lens parallax for range finding, and ending. A weighted averaging method is employed for the calculation of a final parallax, and can eliminate a range finding error better. A scene is observed from a plurality of viewpoints, so as to obtain perception images at different visual angles. The position deviation among pixels of the images is calculated through a triangular measurement principle, so as to obtain the three-dimensional information of the scene. The multi-lens stereoscopic vision parallax method is given on the basis of a stereoscopic parallel vision model. A parallax calculation method of the stereoscopic parallel vision model is employed for the mutual parallax calculation of a plurality of lenses, thereby obtaining a more precise imaging parallax, and enabling the depth information of object points in three-dimensional reconstruction to be more precise. The plurality of lenses can rotate around a circle center in each distribution mode at a proportion with the unchanged relative distance, or can be arranged in a turnover manner along the axis of a line in a plane.

Description

The computational methods of many lens stereos vision parallax

Technical field

The invention belongs to computer vision field, especially relate to the computational methods of a kind of many lens stereos vision parallax.

Background technology

Stereoscopic vision is an important branch in computer vision, and stereoscopic vision is most important in computer passive ranging method Perceived distance technology, it directly simulates human vision and processes the mode of scenery, can measure scenery the most neatly Steric information, its effect is that other computer vision methods is not caned substituted.Research to it, either from vision physiological Angle still all have in engineer applied and be of great significance.The ultimate principle of stereoscopic vision is to regard from two (or multiple) Point observes same scenery, to obtain the perceptual image under different visual angles, calculates the position between image pixel by principle of triangulation Putting deviation (i.e. parallax) and obtain the three-dimensional information of scenery, this process is similar with the three-dimensional perception of human vision.

During having been supplied in the vision-based detection of product on production line, automatically monitoring based on two-dimensional image vision detection technology, but two Two-dimensional projection's features such as the relative position of target, form, labelling can only be differentiated and detect by dimension vision-based detection, are limited The single view projection vision-based detection of local, it is impossible to three-dimensional feature and surface parameter to target object carry out high-precision measurement and three Dimension form identification, numerical DC speed and intelligence during therefore two-dimensional visual detection technique can not meet the most far away modern industry production development Can manufacture and the needs of detection.

Three-dimensional reconstruction based on computer vision, refers to by two width or two-dimensional images to recover the geometry of space object Information.The research of the three-dimensional reconstruction system being then based on binocular stereo vision is developed, and stereo visual system is generally divided into solid Parallel visual system and three-dimensional convergence visual system.If the optical axis of two video cameras is arranged in parallel, the most three-dimensional parallel vision system System；If two optical axises intersect, converge on extraterrestrial target, be referred to as solid and converge visual system.One complete stereo visual system Image Acquisition, Image semantic classification, camera calibration, Stereo matching and five steps of three-dimensional reconstruction can be divided into.

Present inventor's thunder accumulate and strange wait (Lei Yunqi, Song Xiaobing, Yuan Meiling, etc. a kind of face Stereo matching in binocular vision and Parallax calculation method [J]. Xiamen University's journal: natural science edition, 2009,48 (1): 36-41) report in binocular vision A kind of face Stereo matching and parallax calculation method.

Summary of the invention

Present invention aims to deficiency present in the Image Acquisition in existing stereo visual system and disparity computation, carry Computational methods for a kind of many lens stereos vision parallax.

The present invention includes:

One initialized step；

One step obtaining the range finding of all binocular parallaxs；

One step obtaining many camera lenses vision range finding；

One end step.

Described many camera lenses are 3 camera lenses, 4 camera lenses, 5 camera lenses, 6 camera lenses, 7 camera lenses, 8 camera lenses, 9 camera lenses and 10～2500 Camera lens.

Described 3 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.13 camera lenses lay mode

3 camera lenses lay mode and include equilateral triangle, isosceles triangle, arbitrary triangle or the most equidistantly lay；

1.2 initialize all parametric variables

Make each two video camera in three video cameras on same plane that object P shooting is obtained two dimensional image, put down according to solid Row camera system vision mode and its computational methods, there is object P parallax on two image surfaces between any two in three camera lenses, It is denoted as l respectively₁–l₂, l₂–l₃, l₁–l₃, it illustrates P position difference of imaging point in the become image of each two video camera； If d represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₁With S₃Found range from for d₁₃；

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)

d₂₃=(L₂₃*f)/(l₂-l₃)

d₁₃=(L₁₃*f)/(l₁-l₃)

3,3 camera lens vision range finding steps are obtained

3.1 for equilateral triangle, L₁₂=L₂₃=L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.2 for isosceles triangle,The most finally range finding d is:

$d=\frac{2}{5}{d}_{12}+\frac{1}{5}{d}_{23}+\frac{2}{5}{d}_{13}$

3.3 for arbitrary triangle, L₂₃≠L₁₂≠L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.4 for the most equidistantly laying, 2L₂₃=2L₁₂=L₁₃, the most finally range finding d is:

$d=\frac{d_{12}+d_{23}}{4}+d_{13}/2$

4, end step

Calculate 3 camera lens vision range finding results according to as above formula, result is exported.

Described 4 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.14 camera lenses lay mode.4 camera lenses lay mode and include square；Rectangle；3 camera lens composition equilateral triangles, position On same circle, another camera lens is in the center of circle；Circle arbitrarily lays；3 camera lens composition equilateral triangles, are positioned at same circle On, another camera lens is in the centre of following 2 the camera lens lines of equilateral triangle；The most equidistantly lay；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄Represent the optical center position of four video cameras, L respectively₁₃, L₁₂, L₂₃, L₁₄, L₂₄, L₃₄Respectively Represent that the distance between every pair of photographic head optical center, the focal length of four video cameras are f, make four video cameras on same plane In each two video camera to object P shooting obtain two dimensional image, according to three-dimensional parallel camera system vision mode and its calculate Method, there is object P parallax on two image surfaces, is denoted as l respectively in four camera lenses between any two₁–l₂, l₂–l₃, l₁–l₃, l₁–l₄, l₂–l₄, l₃–l₄, it illustrates P position difference of imaging point in the become image of each two video camera.D represents object The distance in P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₂With S₄Found range from for d₂₄, S₃With S₄Found range from for d₃₄。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₃₄=(L₃₄*f)/(l₃-l₄)

3,4 camera lens vision range finding steps are obtained

3.1 for square, L₁₂=L₂₃=L₃₄=L₁₄, the most finally range finding d is:

$d=\frac{1}{4+2\sqrt{2}}{d}_{12}+\frac{1}{4+2\sqrt{2}}{d}_{23}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{13}+\frac{1}{4+2\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{24}+\frac{1}{4+2\sqrt{2}}{d}_{34}$

3.2 for rectangle,The most finally range finding d is:

$d=\frac{1}{6+2\sqrt{5}}{d}_{12}+\frac{1}{3+\sqrt{5}}{d}_{23}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{13}+\frac{1}{3+\sqrt{5}}{d}_{14}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{24}+\frac{1}{6+2\sqrt{5}}{d}_{34}$

3.3, for 3 camera lenses composition equilateral triangles, are positioned on same circle, another camera lens in the center of circle, L₂₃=L₁₂=L₁₃, The most finally range finding d is:

$d=\frac{1}{3+\sqrt{3}}{d}_{12}+\frac{1}{3+\sqrt{3}}{d}_{23}+\frac{1}{3+\sqrt{3}}{d}_{13}+\frac{1}{3+3\sqrt{3}}{d}_{14}+\frac{1}{3+3\sqrt{3}}{d}_{24}+\frac{1}{3+3\sqrt{3}}{d}_{34}$

3.4 for arbitrarily laying on circle, and camera lens distance between any two is different, and the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃+d₁₄+d₂₄+d₃₄)/6

3.5, for 3 camera lens composition equilateral triangles, are positioned on same circle, following 2 at equilateral triangle, another camera lens The centre of camera lens line；Or the most equidistantly lay；First obtain each pair of lens pitch from sum: L=L₁₃+L₁₂+L₂₃+L₁₄+L₂₄+L₃₄, the most finally range finding d is weighted average summation:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 3, and j is from 2 to 4, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Described 5 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.15 camera lenses lay mode

The mode that lays of 5 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another camera lens outer at circle and with just 2 camera lens composition isosceles triangles on square one side；4 camera lens composition squares, are positioned on same circle, another camera lens cloth It is placed on the center of circle；Regular pentagon；The most equidistantly lay；4 camera lens composition squares, are positioned on same circle, another Individual camera lens is in the centre of following 2 the camera lens lines of square；4 lens group rectangularities, another camera lens is in rectangle following 2 The centre of individual camera lens line；Camera lens divides 2 arrangements to put, and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅Represent the optical center position of five video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₁₅, L₁₃, L₁₄, L₂₄, L₂₅, L₃₅Represent that the distance between every pair of camera optics center, the focal length of five video cameras are f respectively.Order is same The each two video camera in five video cameras in one plane obtains two dimensional image to object P shooting, according to the parallel shooting of solid is System vision mode and its computational methods, there is object P parallax on two image surfaces, remember respectively in five camera lenses between any two Make l₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₁–l₅, l₁–l₃, l₁–l₄, l₂–l₄, l₂–l₅, l₃–l₅It illustrates P often The position difference of imaging point in two become images of video camera.D represents the distance in object P anomaly face, then S₁With S₂Found range From for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₁ With S₅Found range from for d₁₅, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₂With S₄Found range from For d₂₄, S₂With S₅Found range from for d₂₅, S₃With S₅Found range from for d₃₅。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₃=(L₁₃*f)/(L₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₃₅=(L₃₅*f)/(l₃-l₅)

3,5 camera lens vision range finding steps are obtained

3.1, for 4 camera lenses composition square, are positioned on same circle, another camera lens outer at circle and with 2 of square Camera lens composition isosceles triangle, L₁₂=L₂₃=L₃₄=L₁₄,The most finally range finding d is:

$\begin{array}{c}d=\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{12}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{23}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{34}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{45}+\frac{\sqrt{2}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{15}\\ +\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{13}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{14}+\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{24}+\frac{\sqrt{2}}{4+3\sqrt{2}+2\sqrt{10}}{d}_{25}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{35}\end{array}$

3.2, for 4 camera lens composition squares, are positioned on same circle, and another camera lens cloth is placed on the center of circle, L₁₂=L₃₄=L₁₄=L₂₃, the most finally range finding d is:

$\begin{array}{c}d=\frac{1}{4+4\sqrt{2}}{d}_{12}+\frac{1}{4+4\sqrt{2}}{d}_{23}+\frac{1}{4+4\sqrt{2}}{d}_{34}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{45}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{15}\\ +\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{13}+\frac{1}{4+4\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{24}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{25}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{35}\end{array}$

3.3 for regular pentagon, L₁₂=L₂₃=L₃₄=L₄₅=L₁₅, the most finally range finding d is:

$\begin{array}{c}d=\frac{1}{5+5\mathrm{\α}}{d}_{12}+\frac{1}{5+5\mathrm{\α}}{d}_{23}+\frac{1}{5+5\mathrm{\α}}{d}_{34}+\frac{1}{5+5\mathrm{\α}}{d}_{45}+\frac{1}{5+5\mathrm{\α}}{d}_{15}\\ +\frac{\mathrm{\α}}{5+5\mathrm{\α}}{d}_{13}+\frac{\mathrm{\α}}{5+5\mathrm{\α}}{d}_{14}+\frac{\mathrm{\α}}{5+5\mathrm{\α}}{d}_{24}+\frac{\mathrm{\α}}{5+5\mathrm{\α}}{d}_{25}+\frac{\mathrm{\α}}{5+5\mathrm{\α}}{d}_{35}\end{array}$

Wherein, a represents d₁₃、d₁₄、d₂₄、d₂₅、d₃₅Binocular range measurement at the ratio of all two camera lens binocular range measurement summations Example, takes any value (it is said that in general, value is more than or equal to 1) between 0.6 to 2.4, desirable incremental step a length of 0.1.

3.4 for the most equidistantly laying；Or 4 camera lens composition squares, it is positioned on same circle, another camera lens Centre at following 2 the camera lens lines of square；Or 4 lens group rectangularities, following 2 in rectangle, another camera lens The centre of camera lens line；Or camera lens divides 2 arrangements to put, adjacent camera lens all forms equilateral triangle；First obtain each pair of lens pitch from Sum: L=L₁₂+L₂₃+L₃₄+L₄₅+L₁₅+L₁₃+L₁₄+L₂₄+L₂₅+L₃₅, the most finally range finding d is weighted average summation:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 4, and j is from 2 to 5, and i is more than j.

4, end step

Calculate 5 camera lens vision range finding results according to as above formula, result is exported.

Described 6 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.16 camera lenses lay mode

The mode that lays of 6 camera lenses includes: 4 camera lens composition squares, is positioned on same circle, and another 2 camera lenses are outer and difference at circle 2 camera lenses adjacent with square both sides and the center of circle respectively form 2 squares；4 camera lens composition squares, are positioned at same circle On, a camera lens cloth is placed on the center of circle, and outer at circle and with square below 2 camera lenses of another camera lens form isosceles Triangle；5 camera lens composition regular pentagons, are positioned on same circle, and another 1 camera lens cloth is placed on the center of circle；Equilateral hexagon, position On same circle；The most equidistantly lay；4 camera lens composition squares, are positioned on same circle, and another 2 camera lenses are each The centre of following 2 camera lens lines on square；6 camera lens rectangles；3 camera lens one big equilateral triangles in outside of composition, Another 3 cloth are placed on the centre of big equilateral triangle each edge, the little equilateral triangle that composition stands upside down；2 arrangements are divided to put, adjacent Camera lens all forms equilateral triangle.

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆Represent the optical center position of six video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₁₆, L₁₃, L₁₄, L₁₅, L₂₄, L₂₅, L₂₆, L₃₅, L₃₆, L₄₆Represent the distance between six camera optics centers respectively, The focal length of six video cameras is f.Make each two video camera in six video cameras on same plane that object P is shot acquisition two Dimension image, according to three-dimensional parallel camera system vision mode and its computational methods, there is object P and exist in six camera lenses between any two Parallax on two image surfaces, is denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₁–l₆, l₁–l₃, l₁–l₄, l₁–l₅, l₂–l₄, l₂–l₅, l₂–l₆, l₃–l₅, l₃–l₆, l₄–l₆It illustrates P and becomes in the become image of each two video camera The position difference of picture point；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range From for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₁ With S₆Found range from for d₁₆, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from For d₁₅, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₄With S₆Found range from for d₄₆；

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₁₆==(L₁₆*f)/(l₁-l₆)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₄₆=(L₄₆*f)/(l₄-l₆)

3,6 camera lens vision range finding steps are obtained

Difference according to 6 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄+ L₄₅+L₅₆+L₁₆+L₁₃+L₁₄+L₁₅+L₂₄+L₂₅+L₂₆+L₃₅+L₃₆+L₄₆, the most finally range finding d is weighted average summation:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 5, and j is from 2 to 6, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Described 7 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.1 7 camera lenses lay mode

The summit of 7 camera lenses lays mode and includes: 4 camera lens composition squares, is positioned on same circle, and another 3 camera lenses are outer also at circle It is adjacent 2 camera lenses on square limit and the center of circle respectively forms 3 squares；4 camera lens composition squares, are positioned at same On circle, a camera lens cloth is placed on the center of circle, another 2 outer at circle and be adjacent 2 camera lenses on square limit and the center of circle respectively forms 2 squares；4 camera lenses composition square, is positioned on same circle, and a camera lens cloth is placed on the center of circle, another 2 outer at circle and with 2 camera lenses on its adjacent square limit respectively form 2 isosceles triangles；6 camera lens composition equilateral hexagon, are positioned at same On circle, a camera lens cloth is placed on the center of circle；Form equilateral heptagon, be positioned on same circle；The most equidistantly lay；4 Individual camera lens composition square, is positioned on same circle, another 2 camera lenses respectively centre of following 2 camera lens lines on square, 1 Individual camera lens is on the right and 2 camera lenses being adjacent form equilateral or isosceles triangles；4 camera lens composition squares, are positioned at same On one circle, another 2 camera lenses respectively centre of following 2 camera lens lines on square, another 1 camera lens is on top and and the upper left corner Equilateral or isosceles triangle is formed with the 2 of the upper right corner camera lenses；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle； Divide 2 arrangements to put and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆, S₇Represent the optical center position of seven video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₆₇, L₁₇, L₁₃, L₁₄, L₁₅, L₁₆, L₂₄, L₂₅, L₂₆, L₂₇, L₃₅, L₃₆, L₃₇, L₄₆, L₄₇, L₅₇Represent respectively Distance between every pair of camera optics center, the focal length of seven video cameras is f, makes in seven video cameras on same plane Each two video camera obtains two dimensional image to object P shooting, according to three-dimensional parallel camera system vision mode and its computational methods, There is object P parallax on two image surfaces in seven camera lenses, is denoted as l between any two respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₁–l₇, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₃–l₅, l₃–l₆, l₃–l₇, l₄–l₆, l₄–l₇, l₅–l₇, it illustrates P position difference of imaging point in the become image of each two video camera；d Represent the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄ Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from for d₆₇, S₁With S₇Found range from for d₁₇, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range From for d₁₅, S₁With S₆Found range from for d₁₆, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂ With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from For d₃₆, S₃With S₇Found range from for d₃₇, S₄With S₆Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₅With S₇Found range from for d₅₇；

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₁₇=(L₁₇*f)/(l₁-l₇): d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)

d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*f)/(l₁-l₆)；d₂₄=(L₂₄*f)/(l₂-L₄)

d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)；d₅₇=(L₅₇*f)/(l₅-l₇)

3,7 camera lens vision range finding steps are obtained

Difference according to 7 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄ +L₄₅+L₅₆+L₆₇+L₁₇+L₁₃+L₁₄+L₁₅+L₁₆+L₂₄+L₂₅+L₂₆+L₂₇+L₃₅+L₃₆+L₃₇+L₄₆+L₄₇+L₅₇, the most finally find range d Sue for peace for weighted average:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 6, and j is from 2 to 7, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Described 8 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.18 camera lenses lay mode

The mode that lays of 8 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another 3 camera lenses outer at circle and and its 2 camera lenses and the center of circle on adjacent square limit respectively form 3 squares, and another 1 camera lens lays in Far Left symmetry；4 Camera lens composition square, is positioned on same circle, and another 3 camera lenses are outer at circle and are adjacent 2 camera lenses on square limit and circle The heart respectively forms 3 squares, and another 1 camera lens cloth is placed on the center of circle；7 camera lenses form equilateral heptagon, are positioned on same circle, Another camera lens cloth is placed on the center of circle；8 camera lenses form equilateral octagon, are positioned on same circle；The most equidistantly lay； Divide 2 arrangements to put and adjacent camera lens all forms equilateral triangle；Divide 2 arrangements to put and adjacent camera lens all forms square；Divide 3 Arrangement is put and adjacent camera lens all forms square；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈Represent the optical center position of eight video cameras respectively, represent respectively with Lij (wherein i is from 1 to 7, and j is from 2 to 8, and i is more than j), Jiao of eight video cameras for distance between every pair of camera optics center Away from being f, each two video camera in eight video cameras on same plane is made object P shooting to be obtained two dimensional image, according to vertical Body parallel camera system vision mode and its computational methods, there is object P on two image surfaces in eight camera lenses between any two Parallax, is denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇–l₈, l₁–l₈, l₁–l₃, l₁–l₄, l₁– l₅, l₁–l₆, l₁–l₇, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₄–l₆, l₄–l₇, l₄–l₈, l₅–l₇, l₅–l₈, l₆–l₈, it illustrates P alternate position spike of imaging point in the become image of each two video camera Different；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃ With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from For d₆₇, S₇With S₈Found range from for d₇₈, S₁With S₈Found range from for d₁₈, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₁With S₇Found range from for d₁₇, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇ Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₄With S₆Found range from for d₄₆, S₄With S₇Found range From for d₄₇, S₄With S₈Found range from for d₄₈, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₆ With S₈Found range from for d₆₈；

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₁₈=(L₁₈*f)/(l₁-l₈)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*f)/(l₁-l₆)

d₁₇=(L₁₇*f)/(l₁-l₇)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)

d₄₈=(L₄₈*f)/(l₄-l₈)；d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)

d₆₈=(L₆₈*f)/(l₆-l₈)

3,8 camera lens vision range finding steps are obtained

Difference according to 8 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, and the most finally range finding d is for adding Weight average is sued for peace:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 7, and j is from 2 to 8, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Described 9 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.19 camera lenses lay mode

The mode that lays of 9 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another 3 camera lenses outer at circle and and its 2 camera lenses and the center of circle on adjacent square limit respectively form 3 squares, and 1 camera lens lays in Far Left symmetry, 1 camera lens Cloth is placed on the center of circle；8 camera lenses form equilateral octagon, and 1 camera lens cloth is placed on the center of circle；3 rows are divided equidistantly to lay, adjacent mirror Head all forms square；8 camera lenses divide 2 rows equidistantly to lay, and all form square, another 1 camera lens on the right and with its phase 2 adjacent camera lenses form equilateral or isosceles triangle；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle；Divide 2 rows Lay and adjacent camera lens all forms equilateral triangle；The most equidistantly lay；

1.2 initialize all parametric variables

S1, S2, S3, S4, S5, S6, S7, S8, S9 represent the optical center position of nine video cameras respectively, with Lij respectively (wherein i is from 1 to 8, and j is from 2 to 9, and i is more than j), nine video cameras to represent the distance between every pair of camera optics center Focal length be f, make each two video camera in nine video cameras on same plane that object P shooting is obtained two dimensional image, root According to three-dimensional parallel camera system vision mode and its computational methods, there is object P between any two at two image surfaces in nine camera lenses On parallax, be denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇–l₈, l₈–l₉, l₁–l₉, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₁–l₇, l₁–l₈, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₂–l₉, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₃–l₉, l₄–l₆, l₄–l₇, l₄–l₈, l₄–l₉, l₅–l₇, l₅–l₈, l₅–l₉, l₆–l₈, l₆–l₉, l₇–l₉ It illustrates P position difference of imaging point in the become image of each two video camera；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range From for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from for d₆₇, S₇With S₈Found range from for d₇₈, S₈ With S₉Found range from for d₈₉, S₁With S₉Found range from for d₁₉, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from For d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₁With S₇Found range from for d₁₇, S₁With S₈Found range from for d₁₈, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₂With S₉Found range from for d₂₉, S₃With S₅ Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₃With S₉Found range from for d₃₉, S₄With S₆Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₄With S₈Found range From for d₄₈, S₄With S₉Found range from for d₄₉, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₅ With S₉Found range from for d₅₉, S₆With S₈Found range from for d₆₈, S₆With S₉Found range from for d₆₉, S₇With S₉Found range from For d₇₉；

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₈₉=(L₈₉*f)/(l₈-l₉)；d₁₉=(L₁₉*f)/(l₁-l₉)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₁₆=(L₁₆*f)/(l₁-l₆)；d₁₇=(L₁₇*f)/(l₁-l₇)；d₁₈=(L₁₈*f)/(l₁-l₈)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)；d₂₉=(L₂₉*f)/(l₂-l₉)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₃₉=(L₃₉*f)/(l₃-l₉)；d₄₆=(L₄₆*f)/(l₄-l₆)

d₄₇=(L₄₇*f)/(l₄-l₇)；d₄₈=(L₄₈*f)/(l₄-l₈)；d₄₉=(L₄₉*f)/(l₄-l₉)

d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)；d₅₉=(L₅₉*f)/(l₅-l₉)

d₆₈=(L₆₈*f)/(l₆-l₈)；d₆₉=(L₆₉*f)/(l₆-l₉)；d₇₉=(L₇₉*f)/(l₇-l₉)

3,9 camera lens vision range finding steps are obtained

Difference according to 9 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, and the most finally range finding d is for adding Weight average is sued for peace:

$d=\mathrm{\Σ}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1～8, and j is from 2～9, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Described 10～2500 lens stereo vision parallax calculation methods comprise the following steps:

1, initialization step

1.1 camera lenses lay mode

The arrangement mode of 10～2500 camera lenses obtains by the following method:

1) note camera lens number is N, 10≤N≤2500；

2) arrangement maximum number of lines M:M is calculated²≤N<(M+1)², M is an integer determining of corresponding N and 3≤M≤50；

3) selected line number H to be arranged, 1≤H≤M；

3.1) if H=1, the most each camera lens constitutes straight line and equidistant arrangement；

3.2) if H=2, in mode below select one:

The most upper and lower 2 row are parallel, and adjacent nearest every 4 camera lenses constitute square；

The most upper and lower 2 row are parallel, and adjacent nearest every 3 camera lenses constitute equilateral triangle；

If 3.2.3 N is odd number, then last camera lens 2 camera lenses rightmost with upper and lower 2 row constitute equilateral triangle Shape, or direct equidistant cloth is placed on upstream or downstream；

3.3) if H >=3, parallel and equidistant between each row:

3.3.1 according to H, minimum camera lens number I:H × I≤N < H × (I+1) that often row lays is determined；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 4 camera lenses constitute square, or take 3.3.3 Mode lays；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 3 camera lenses constitute equilateral triangle；

3.3.4 K the camera lens (1≤K < H) having more, the most often they cloth are placed on the first row to middle the prolonging of line k by row one On long line, and rightmost 2 camera lenses of 2 row upper and lower with it constitute equilateral triangle；Or often row one is by they equidistant cloth Being placed on the first row to the extended line of line k, 4 camera lenses being adjacent constitute square.

1.2 initialize all parametric variables

Under the mode that lays of 10～2500 camera lenses, Si (i is from 1 to N) represents the optical center position of each video camera, divides with Lij Do not represent distance between every pair of camera optics center (wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j), each The focal length of video camera is f, makes each two video camera on same plane that object P shooting is obtained two dimensional image, puts down according to solid There is object P between any two on two image surfaces in row camera system vision mode and its computational methods, camera lens i and camera lens j Parallax, be denoted as (li lj), it illustrates P position difference of imaging point in the become image of each two video camera.D represents object The distance in P anomaly face, then Si Yu Sj is found range from for dij.

2, all binocular parallax ranging step are obtained

Binocular parallax range finding result of calculation between each two camera lens Si and camera lens Sj is:

d_ij=(L_ij*f)/(l_i-l_j)

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j.

3,10～2500 camera lens vision range finding steps are obtained

Lay mode according to 10～2500 each differences of camera lens, first obtain each pair of lens pitch from sum: L=∑ Lij, the most finally survey It is weighted average summation away from d:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j.

4, end step

Calculate 10～2500 camera lens vision range finding results according to as above formula, result is exported.

The present invention, from the same scenery of multiple viewing point, to obtain the perceptual image under different visual angles, passes through principle of triangulation Calculate the position deviation (i.e. parallax) between image pixel and obtain the three-dimensional information of scenery.At base based on three-dimensional parallel vision mode On plinth, provide the shooting of many lens stereos vision and the computational methods of parallax thereof.Utilize three-dimensional parallel in the environment of multi-lens imaging Many camera lenses are carried out disparity computation, it is thus achieved that the most accurate image disparity, make by the parallax calculation method of vision mode each other Obtain object point depth information in three-dimensional reconstruction the most accurate.

Accompanying drawing explanation

Fig. 1 is three-dimensional parallel camera system vision mode schematic diagram.

Fig. 2 is that 3 camera lenses lay mode (3 camera lens-1).

Fig. 3 is that 3 camera lenses lay mode (3 camera lens-2).

Fig. 4 is that 3 camera lenses lay mode (3 camera lens-3).

Fig. 5 is that 3 camera lenses lay mode (3 camera lens-4).

Fig. 6 is that 4 camera lenses lay mode (4 camera lens-1).

Fig. 7 is that 4 camera lenses lay mode (4 camera lens-2).

Fig. 8 is that 4 camera lenses lay mode (4 camera lens-3).

Fig. 9 is that 4 camera lenses lay mode (4 camera lens-4).

Figure 10 is that 4 camera lenses lay mode (4 camera lens-5).

Figure 11 is that 4 camera lenses lay mode (4 camera lens-6).

Figure 12 is that 5 camera lenses lay mode (5 camera lens-1).

Figure 13 is that 5 camera lenses lay mode (5 camera lens-2).

Figure 14 is that 5 camera lenses lay mode (5 camera lens-3).

Figure 15 is that 5 camera lenses lay mode (5 camera lens-4).

Figure 16 is that 5 camera lenses lay mode (5 camera lens-5).

Figure 17 is that 5 camera lenses lay mode (5 camera lens-6).

Figure 18 is that 5 camera lenses lay mode (5 camera lens-7).

Figure 19 is that 6 camera lenses lay mode (6 camera lens-1).

Figure 20 is that 6 camera lenses lay mode (6 camera lens-2).

Figure 21 is that 6 camera lenses lay mode (6 camera lens-3).

Figure 22 is that 6 camera lenses lay mode (6 camera lens-4).

Figure 23 is that 6 camera lenses lay mode (6 camera lens-5).

Figure 24 is that 6 camera lenses lay mode (6 camera lens-6).

Figure 25 is that 6 camera lenses lay mode (6 camera lens-7).

Figure 26 is that 6 camera lenses lay mode (6 camera lens-8).

Figure 27 is that 6 camera lenses lay mode (6 camera lens-9).

Figure 28 is that 7 camera lenses lay mode (7 camera lens-1).

Figure 29 is that 7 camera lenses lay mode (7 camera lens-2).

Figure 30 is that 7 camera lenses lay mode (7 camera lens-3).

Figure 31 is that 7 camera lenses lay mode (7 camera lens-4).

Figure 32 is that 7 camera lenses lay mode (7 camera lens-5).

Figure 33 is that 7 camera lenses lay mode (7 camera lens-6).

Figure 34 is that 7 camera lenses lay mode (7 camera lens-7).

Figure 35 is that 7 camera lenses lay mode (7 camera lens-8).

Figure 36 is that 7 camera lenses lay mode (7 camera lens-9).

Figure 37 is that 7 camera lenses lay mode (7 camera lens-10).

Figure 38 is that 8 camera lenses lay mode (8 camera lens-1).

Figure 39 is that 8 camera lenses lay mode (8 camera lens-2).

Figure 40 is that 8 camera lenses lay mode (8 camera lens-3).

Figure 41 is that 8 camera lenses lay mode (8 camera lens-4).

Figure 42 is that 8 camera lenses lay mode (8 camera lens-5).

Figure 43 is that 8 camera lenses lay mode (8 camera lens-6).

Figure 44 is that 8 camera lenses lay mode (8 camera lens-7).

Figure 45 is that 8 camera lenses lay mode (8 camera lens-8).

Figure 46 is that 8 camera lenses lay mode (8 camera lens-9).

Figure 47 is that 9 camera lenses lay mode (9 camera lens-1).

Figure 48 is that 9 camera lenses lay mode (9 camera lens-2).

Figure 49 is that 9 camera lenses lay mode (9 camera lens-3).

Figure 50 is that 9 camera lenses lay mode (9 camera lens-4).

Figure 51 is that 9 camera lenses lay mode (9 camera lens-5).

Figure 52 is that 9 camera lenses lay mode (9 camera lens-6).

Figure 53 is that 9 camera lenses lay mode (9 camera lens-7).

Detailed description of the invention

Following example will the present invention is further illustrated in conjunction with accompanying drawing.

On the basis of based on three-dimensional parallel vision mode, the shooting of many lens stereos vision and the computational methods of parallax thereof are proposed, Utilize the parallax calculation method of three-dimensional parallel vision mode in the environment of multi-lens imaging, many camera lenses are carried out each other parallaxometer Calculate, it is thus achieved that the most accurate image disparity so that in three-dimensional reconstruction, object point depth information is the most accurate.

The present invention proposes the shooting of many lens stereos vision and the computational methods of parallax, the disparity computation of many lens stereos vision mode Method binocular stereo vision parallax calculation method based on three-dimensional parallel camera system.

As it is shown in figure 1, wherein, P represents subject, C_lAnd C_rRepresent camera lens.If C_lAnd C_rIt is respectively a left side The optical center position of right two cameras, C_lAnd C_rBetween distance be b, the focal length of two cameras is f.P_lAnd P_rIt is referred to as sky Between put the P corresponding subpoint in the image plane of left and right, P and C_lAnd C_rDistance between line is d.Cross C_lAnd C_rRespectively to View plane makees vertical line, and intersection point is respectively A_lAnd A_r, cross P and make vertical line to view plane, order | A_lP_l|=l_a, | A_rP_r|=l_b, | P_rB |= A, then learnt equation below by the relation between similar triangles:

(d-f)/d=a/ (a+l_b) formula 1

(d-f)/d=(b-l_a+l_b+a)/(a+l_b+ b) formula 2

Equation below 3 is pushed away to obtain by formula 1 and 2:

a/(a+l_b)=(b-l_a+l_b+a)/(a+l_b+ b)=1-l_a/(b+l_b+ a) formula 3

Thus have equation below 4:

A=(b*l_a)/(l_a-l_b)-l_bFormula 4

Formula 4 is brought in formula 1 and obtains formula 5:

D=f* (a+l_b)/l_b=(b*f)/(l_a-l_b) formula 5

By formula 5 it can be seen that distance d and b, f and l_a-l_bRelevant.l_a-l_bIt is referred to as some P on the image surface of two, left and right Parallax, it illustrates P point position difference of imaging point in left images.Owing to b, f are known, solid to be realized Vision range finding, most critical seeks to try to achieve parallax l_a-l_b。

Currently known binocular parallel camera system model disparity computation mode, the present invention takes polygon vertex to lay the side of many camera lenses Formula forms the stereoscopic vision shooting of many camera lenses, calculates new parallax, to realize stereopsis range finding.

The shooting of many lens stereos vision that polygon vertex lays include 3 lens stereo visions shootings, 4 lens stereo visions shootings, 5 lens stereo visions shooting, 6 lens stereo visions shooting, 7 lens stereo visions shooting, 8 lens stereo visions shooting, 9 Lens stereo vision images, and the stereoscopic vision shooting of more camera lens.Many camera lenses are generally aligned in the same plane, with different summit cloth The mode of putting realizes stereopsis ranging process accurate to object.The present invention represents what each optical center was positioned at dashed circle Circular flat.Every kind of many camera lenses laid in mode can rotate around the center of circle according to the ratio that relative distance is constant；Also can be according to flat On face, straight line is that axle upset lays camera lens.Connect with the dotted line between two camera lenses and represent the distance between two lens optical centers.

Every kind of many lens stereos vision shooting of the present invention and the computational methods of parallax thereof by initialization step, obtain all binocular visions Differ from ranging step, obtain many camera lenses vision range finding step, end step composition.Wherein have employed weighting when calculating final parallax Average method, can preferably eliminate range error.

One, 3 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.1 3 camera lenses lay mode.The mode that lays of 3 camera lenses has 4 kinds, the camera lens composition equilateral triangle of Fig. 2, is positioned at same On circle；The camera lens composition isosceles triangle of Fig. 3, is positioned on same circle；The camera lens composition arbitrary triangle of Fig. 4, is positioned at same On circle；The camera lens of Fig. 5 the most equidistantly lays.

1.2 initialize all parametric variables

Lay in schematic diagram 4 kinds of 3 camera lenses, S₁, S₂, S₃Represent the optical center position of three video cameras, L respectively₁₃, L₁₂, L₂₃Representing the distance between corresponding every pair of camera optics center respectively, the focal length of three video cameras is f.Make same flat The each two video camera in three video cameras on face obtains two dimensional image to object P shooting, regards according to three-dimensional parallel camera system Vision model and its computational methods, there is object P parallax on two image surfaces, be denoted as l respectively in three camera lenses between any two₁– l₂, l₂–l₃, l₁–l₃, it illustrates P position difference of imaging point in the become image of each two video camera.D represents object P The distance in anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₁With S₃Found range from for d₁₃。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)

d₂₃=(L₂₃*f)/(l₂-l₃)

d₁₃=(L₁₃*f)/(l₁-l₃)

3, many camera lenses vision range finding step is obtained

3.1 for Fig. 2, L₁₂=L₂₃=L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.2 for Fig. 3,The most finally range finding d is:

$d=\frac{2}{5}{d}_{12}+\frac{1}{5}{d}_{23}+\frac{2}{5}{d}_{13}$

3.3 for Fig. 4, L₂₃≠L₁₂≠L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.3 for Fig. 5,2L₂₃=2L₁₂=L₁₃, the most finally range finding d is:

$d=\frac{d_{12}+d_{23}}{4}+d_{13}/2$

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Two, 4 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.14 camera lenses lay mode.The mode that lays of 4 camera lenses has 6 kinds.Wherein, the camera lens composition square of Fig. 6, it is positioned at same On one circle；The lens group rectangularity of Fig. 7 and L₂₃=2L₁₂, it is positioned on same circle；3 camera lens composition equilateral triangles of Fig. 8 Shape, is positioned on same circle, and another camera lens is in the center of circle；The camera lens of Fig. 9 arbitrarily lays on circle；3 camera lens compositions of Figure 10 Equilateral triangle, is positioned on same circle, and another camera lens is in the centre of following 2 the camera lens lines of equilateral triangle；The mirror of Figure 11 Head the most equidistantly lays.

1.2 initialize all parametric variables

Lay in schematic diagram 6 kinds of 4 camera lenses, S₁, S₂, S₃, S₄Represent the optical center position of four video cameras respectively, L₁₃, L₁₂, L₂₃, L₁₄, L₂₄, L₃₄Representing the distance between every pair of photographic head optical center respectively, the focal length of four video cameras is equal For f.Make each two video camera in four video cameras on same plane that object P shooting is obtained two dimensional image, put down according to solid Row camera system vision mode and its computational methods, there is object P parallax on two image surfaces between any two in four camera lenses, It is denoted as l respectively₁–l₂, l₂–l₃, l₁–l₃, l₁–l₄, l₂–l₄, l₃–l₄, it illustrates P in the become image of each two video camera The position difference of imaging point.D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Surveyed Distance is d₂₃, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₂With S₄Found range from for d₂₄, S₃ With S₄Found range from for d₃₄。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₃₄=(L₃₄*f)/(l₃-l₄) 3, obtain Take many camera lenses vision range finding step

3.1 for Fig. 6, L₁₂=L₂₃=L₃₄=L₁₄, the most finally range finding d is:

$d=\frac{1}{4+2\sqrt{2}}{d}_{12}+\frac{1}{4+2\sqrt{2}}{d}_{23}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{13}+\frac{1}{4+2\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{24}+\frac{1}{4+2\sqrt{2}}{d}_{34}$

3.2 for Fig. 7,The most finally range finding d is:

$d=\frac{1}{6+2\sqrt{5}}{d}_{12}+\frac{1}{3+\sqrt{5}}{d}_{23}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{13}+\frac{1}{3+\sqrt{5}}{d}_{14}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{24}+\frac{1}{6+2\sqrt{5}}{d}_{34}$

3.3 for Fig. 8, L₂₃=L₁₂=L₁₃, the most finally range finding d is:

$d=\frac{1}{3+\sqrt{3}}{d}_{12}+\frac{1}{3+\sqrt{3}}{d}_{23}+\frac{1}{3+\sqrt{3}}{d}_{13}+\frac{1}{3+3\sqrt{3}}{d}_{14}+\frac{1}{3+3\sqrt{3}}{d}_{24}+\frac{1}{3+3\sqrt{3}}{d}_{34}$

3.4 for Fig. 9, and camera lens distance between any two is different, and the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃+d₁₄+d₂₄+d₃₄)/6

3.5 for Figure 10 or Figure 11, first obtains each pair of lens pitch from sum: L=L₁₃+L₁₂+L₂₃+L₁₄+L₂₄+L₃₄, The most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 3, and j is from 2 to 4, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Three, 5 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.15 camera lenses lay mode.The mode that lays of 5 camera lenses has 7 kinds.Wherein, 4 camera lens composition squares of Figure 12, position On same circle, outer at circle and with square 2 camera lenses of another camera lens form isosceles triangles；4 mirrors of Figure 13 Head composition square, is positioned on same circle, and another camera lens cloth is placed on the center of circle；5 camera lens composition regular pentagons of Figure 14；Figure 5 camera lenses of 15 the most equidistantly lay；4 camera lens composition squares of Figure 16, are positioned on same circle, another Individual camera lens is in the centre of following 2 the camera lens lines of square；4 lens group rectangularities of Figure 17, another camera lens is rectangular The centre of following 2 the camera lens lines of shape；The camera lens of Figure 18 divides 2 arrangements to put, and adjacent camera lens all forms equilateral triangle.

1.2 initialize all parametric variables

Lay in schematic diagram 7 kinds of 5 camera lenses, S₁, S₂, S₃, S₄, S₅Represent the optical center position of five video cameras respectively Put, L₁₂, L₂₃, L₃₄, L₄₅, L₁₅, L₁₃, L₁₄, L₂₄, L₂₅, L₃₅Represent respectively between every pair of camera optics center away from From, the focal length of five video cameras is f.Make each two video camera on same plane five video cameras that object P shooting is obtained Taking two dimensional image, according to three-dimensional parallel camera system vision mode and its computational methods, there is object in five camera lenses between any two P parallax on two image surfaces, is denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₁–l₅, l₁–l₃, l₁–l₄, l₂–l₄, l₂–l₅, l₃–l₅It illustrates P position difference of imaging point in the become image of each two video camera.D represents object P anomaly face Distance, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄ With S₅Found range from for d₄₅, S₁With S₅Found range from for d₁₅, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from For d₁₄, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₃With S₅Found range from for d₃₅。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-L₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(LL₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₃₅=(L₃₅*f)/(l₃-l₅)

3, many camera lenses vision range finding step is obtained

3.1 for Figure 12, L₁₂=L₂₃=L₃₄=L₁₄,The most finally range finding d is:

$\begin{array}{c}d=\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{12}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{23}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{34}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{45}+\frac{\sqrt{2}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{15}\\ +\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{13}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{14}+\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{24}+\frac{\sqrt{2}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{25}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{35}\end{array}$

3.2 for Figure 13, L₁₂=L₃₄=L₁₄=L₂₃, the most finally range finding d is:

$\begin{array}{c}d\frac{1}{4+4\sqrt{2}}{d}_{12}+\frac{1}{4+4\sqrt{2}}{d}_{23}+\frac{1}{4+4\sqrt{2}}{d}_{34}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{45}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{15}\\ +\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{13}+\frac{1}{4+4\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{24}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{25}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{35}\end{array}$

3.3 for Figure 14, L₁₂=L₂₃=L₃₄=L₄₅=L₁₅, the most finally range finding d is:

$\begin{array}{c}d=\frac{1}{5+5\mathrm{\&alpha;}}{d}_{12}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{23}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{34}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{45}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{15}\\ +\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{13}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{14}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{24}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{25}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{35}\end{array}$

Wherein, a represents d₁₃、d₁₄、d₂₄、d₂₅、d₃₅Binocular range measurement at the ratio of all two camera lens binocular range measurement summations Example, takes any value (it is said that in general, value is more than or equal to 1) between 0.6 to 2.4, desirable incremental step a length of 0.1.

3.4, for Figure 15, Figure 16, Figure 17, Figure 18 difference, first obtain each pair of lens pitch from sum: L= L₁₂+L₂₃+L₃₄+L₄₅+L₁₅+L₁₃+L₁₄+L₂₄+L₂₅+L₃₅, the most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 4, and j is from 2 to 5, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Four, 6 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.1 6 camera lenses lay mode.The mode that lays of 6 camera lenses has 9 kinds.Wherein, 4 camera lens composition squares of Figure 19, position On same circle, 2 camera lenses and the center of circle that another 2 camera lenses are outer at circle and adjacent with square both sides respectively respectively form 2 pros Shape；4 camera lens composition squares of Figure 20, are positioned on same circle, and a camera lens cloth is placed on the center of circle, and another camera lens is outside circle And form isosceles triangles with 2 camera lenses on one side of square below；5 camera lens composition regular pentagons of Figure 21, are positioned at On same circle, a camera lens cloth is placed on the center of circle；The camera lens composition equilateral hexagon of Figure 22, is positioned on same circle；The mirror of Figure 23 Head the most equidistantly lays；4 camera lens composition squares of Figure 24, are positioned on same circle, and another 2 camera lenses are respectively just The centre of square following 2 camera lens lines；Adjacent 4 lens group rectangularities of Figure 25；The 3 camera lens compositions one of Figure 26 The big equilateral triangle in individual outside, another 3 cloth are placed on the centre of big equilateral triangle each edge, the little equilateral triangle that composition stands upside down； The camera lens of Figure 27 divides 2 arrangements to put, and adjacent camera lens all forms equilateral triangle.

1.2 initialize all parametric variables

Lay in schematic diagram at 6 camera lenses, S₁, S₂, S₃, S₄, S₅, S₆Represent the optical center position of six video cameras respectively Put, L₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₁₆, L₁₃, L₁₄, L₁₅, L₂₄, L₂₅, L₂₆, L₃₅, L₃₆, L₄₆Represent six respectively Distance between camera optics center, the focal length of six video cameras is f.Make every two in six video cameras on same plane Individual video camera obtains two dimensional image, according to three-dimensional parallel camera system vision mode and its computational methods, six to object P shooting There is object P parallax on two image surfaces in individual camera lens, is denoted as l between any two respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₁–l₆, l₁–l₃, l₁–l₄, l₁–l₅, l₂–l₄, l₂–l₅, l₂–l₆, l₃–l₅, l₃–l₆, l₄–l₆It illustrates P The position difference of imaging point in the become image of each two video camera.D represents the distance in object P anomaly face, then S₁With S₂Institute Find range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₁With S₆Found range from for d₁₆, S₁With S₃Found range from for d₁₃, S₁With S₄Found range From for d₁₄, S₁With S₅Found range from for d₁₅, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂ With S₆Found range from for d₂₆, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₄With S₆Found range from For d₄₆。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₁₆=(L₁₆*f)/(l₁-l₆)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₄₆=(L₄₆*f)/(l₄-l₆)

3, many camera lenses vision range finding step is obtained

Difference according to 6 each schematic diagrams of camera lens lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄+ L₄₅+L₅₆+L₁₆+L₁₃+L₁₄+L₁₅+L₂₄+L₂₅+L₂₆+L₃₅+L₃₆+L₄₆, the most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 5, and j is from 2 to 6, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Five, 7 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.17 camera lenses lay mode.The summit of 7 camera lenses lays mode 10 kinds.Wherein, 4 camera lens compositions of Figure 28 are square Shape, is positioned on same circle, and another 3 camera lenses are outer at circle and are adjacent 2 camera lenses on square limit and the center of circle respectively forms 3 Square；4 camera lenses composition square of Figure 29, is positioned on same circle, and a camera lens cloth is placed on the center of circle, another 2 outside circle And be adjacent square limit on 2 camera lenses and the center of circle respectively form 2 squares；4 camera lens composition squares of Figure 30, Being positioned on same circle, a camera lens cloth is placed on the center of circle, and another 2 and are adjacent each group of 2 camera lenses on square limit outside circle Become 2 isosceles triangles；6 camera lens composition equilateral hexagon of Figure 31, are positioned on same circle, and a camera lens cloth is placed on the center of circle； The camera lens of Figure 32 forms equilateral heptagon, is positioned on same circle；The camera lens of Figure 33 the most equidistantly lays；Figure 34's 4 camera lens composition squares, are positioned on same circle, another 2 camera lenses respectively centre of following 2 camera lens lines on square, 1 Individual camera lens is on the right and 2 camera lenses being adjacent form equilateral or isosceles triangles；6 camera lenses of Figure 35 form same Figure 34, Another camera lens on top and forms equilateral or isosceles triangles with 2 camera lenses in the upper left corner and the upper right corner；The camera lens of Figure 36 divides 3 Arrangement is put and adjacent camera lens all forms equilateral triangle；The camera lens of Figure 37 divides 2 arrangements to put and adjacent camera lens all forms equilateral three Dihedral.

1.2 initialize all parametric variables

Lay in schematic diagram at 7 camera lenses, S₁, S₂, S₃, S₄, S₅, S₆, S₇Represent respectively in the optics of seven video cameras Heart position, L₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₆₇, L₁₇, L₁₃, L₁₄, L₁₅, L₁₆, L₂₄, L₂₅, L₂₆, L₂₇, L₃₅, L₃₆, L₃₇, L₄₆, L₄₇, L₅₇Represent that the distance between every pair of camera optics center, the focal length of seven video cameras are f respectively.Order is same The each two video camera in seven video cameras in one plane obtains two dimensional image to object P shooting, according to the parallel shooting of solid is System vision mode and its computational methods, there is object P parallax on two image surfaces, remember respectively in seven camera lenses between any two Make l₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₁–l₇, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₃–l₅, l₃–l₆, l₃–l₇, l₄–l₆, l₄–l₇, l₅–l₇, it illustrates P and images in each two The position difference of imaging point in the become image of machine.D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range From for d₅₆, S₆With S₇Found range from for d₆₇, S₁With S₇Found range from for d₁₇, S₁With S₃Found range from for d₁₃, S₁ With S₄Found range from for d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₂With S₄Found range from For d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₄With S₆Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₅With S₇Found range from for d₅₇。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(l₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₁₇=(L₁₇*f)/(l₁-l₇)；d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)

d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*/)/(l₁-l₆)；d₂₄=(L₂₄*f)/(l₂-l₄)

d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)；d₅₇=(L₅₇*f)/(l₅-l₇)

3, many camera lenses vision range finding step is obtained

Difference according to 7 each schematic diagrams of camera lens lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄ +L₄₅+L₅₆+L₆₇+L₁₇+L₁₃+L₁₄+L₁₅+L₁₆+L₂₄+L₂₅+L₂₆+L₂₇+L₃₅+L₃₆+L₃₇+L₄₆+L₄₇+L₅₇, the most finally find range d Sue for peace for weighted average:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 6, and j is from 2 to 7, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Six, 8 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.1 8 camera lenses lay mode.The mode that lays of 8 camera lenses has 9 kinds.Wherein, 7 camera lenses of Figure 38 lay mode with 7 mirrors Head-1, another camera lens lays in Far Left symmetry；7 camera lenses of Figure 39 lay mode with 7 camera lens-1, another camera lens cloth It is placed on the center of circle；7 camera lenses of Figure 40 form equilateral heptagon, are positioned on same circle, and another camera lens cloth is placed on the center of circle；Figure The camera lens of 41 forms equilateral octagon, is positioned on same circle；The camera lens of Figure 42 the most equidistantly lays；The mirror of Figure 43 Head point 2 arrangement is put and adjacent camera lens all forms equilateral triangle；The camera lens of Figure 44 divides 2 arrangements to put and adjacent camera lens all forms Square；The camera lens of Figure 45 divides 3 arrangements to put and adjacent camera lens all forms square；The camera lens of Figure 46 divides 3 arrangements to put and phase Adjacent camera lens all forms equilateral triangle.

1.2 initialize all parametric variables

Lay in schematic diagram at 8 camera lenses, S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈Represent the light of eight video cameras respectively Learn center, with Lij represent respectively distance between every pair of camera optics center (wherein i is from 1 to 7, j from 2 to 8, And i is more than j), the focal length of eight video cameras is f.Make each two video camera in eight video cameras on same plane to object P shooting obtains two dimensional image, and according to three-dimensional parallel camera system vision mode and its computational methods, eight camera lenses are between any two There is object P parallax on two image surfaces, be denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇ –l₈, l₁–l₈, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₁–l₇, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₄–l₆, l₄–l₇, l₄–l₈, l₅–l₇, l₅–l₈, l₆–l₈, it illustrates P and images in each two The position difference of imaging point in the become image of machine.D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range From for d₅₆, S₆With S₇Found range from for d₆₇, S₇With S₈Found range from for d₇₈, S₁With S₈Found range from for d₁₈, S₁ With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from For d₁₆, S₁With S₇Found range from for d₁₇, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₄With S₆ Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₄With S₈Found range from for d₄₈, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₆With S₈Found range from for d₆₈。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₁₈=(L₁₈*f)/(l₁-l₈)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*f)/(l₁-l₆)

d₁₇=(L₁₇*f)/(l₁-l₇)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)

d₄₈=(L₄₈*f)/(l₄-l₈)；d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)

d₆₈=(L₆₈*f)/(l₆-l₈)

3, many camera lenses vision range finding step is obtained

Difference according to 8 each schematic diagrams of camera lens lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, the most finally Range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 7, and j is from 2 to 8, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Seven, 9 shooting of lens stereo vision and disparity computation thereof

1, initialization step

1.19 camera lenses lay mode.The mode that lays of 9 camera lenses has 7 kinds.Wherein, 8 camera lenses of Figure 47 lay mode with 8 mirrors Head-1, another camera lens cloth is placed on the center of circle；8 camera lenses of Figure 48 form equilateral octagon, and a camera lens cloth is placed on the center of circle；Figure The camera lens of 49 divides 3 rows equidistantly to lay, and adjacent camera lens all forms square；8 camera lenses of Figure 50 divide the 2 equidistant cloth of row Putting, all form square, another camera lens is on the right and 2 camera lenses being adjacent form equilateral or isosceles triangles；Figure 51 Camera lens divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle；The camera lens of Figure 52 divides 2 arrangements to put and adjacent camera lens is equal Composition equilateral triangle；The camera lens of Figure 53 the most equidistantly lays.

1.2 initialize all parametric variables

Lay in schematic diagram 7 kinds of 9 camera lenses, S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈, S₉Represent nine respectively to take the photograph The optical center position of camera, (wherein i is from 1 to 8, j to represent the distance between every pair of camera optics center respectively with Lij From 2 to 9, and i is more than j), and the focal length of nine video cameras is f.The each two on same plane nine video cameras is made to take the photograph Camera obtains two dimensional image, according to three-dimensional parallel camera system vision mode and its computational methods, nine mirrors to object P shooting There is object P parallax on two image surfaces in head, is denoted as l between any two respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇–l₈, l₈–l₉, l₁–l₉, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₁–l₇, l₁–l₈, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₂–l₉, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₃–l₉, l₄–l₆, l₄–l₇, l₄–l₈, l₄–l₉, l₅–l₇, l₅–l₈, l₅–l₉, l₆–l₈, l₆–l₉, l₇–l₉It illustrates P position difference of imaging point in the become image of each two video camera. D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄ Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from for d₆₇, S₇With S₈Found range from for d₇₈, S₈With S₉Found range from for d₈₉, S₁With S₉Found range from for d₁₉, S₁With S₃Found range From for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₁ With S₇Found range from for d₁₇, S₁With S₈Found range from for d₁₈, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from For d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₂With S₉Found range from for d₂₉, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₃With S₉Found range from for d₃₉, S₄With S₆Found range from for d₄₆, S₄With S₇ Found range from for d₄₇, S₄With S₈Found range from for d₄₈, S₄With S₉Found range from for d₄₉, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₅With S₉Found range from for d₅₉, S₆With S₈Found range from for d₆₈, S₆With S₉Found range From for d₆₉, S₇With S₉Found range from for d₇₉。

2, all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₈₉=(L₈₉*f)/(l₈-l₉)；d₁₉=(L₁₉*f)/(l₁-l₉)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₁₆=(L₁₆*f)/(l₁-l₆)；d₁₇=(L₁₇*f)/(l₁-l₇)；d₁₈=(L₁₈*f)/(l₁-l₈)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)；d₂₉=(L₂₉*f)/(l₂-l₉)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₃₉=(L₃₉*f)/(l₃-l₉)；d₄₆=(L₄₆*f)/(l₄-l₆)

d₄₇=(L₄₇*f)/(l₄-l₇)；d₄₈=(L₄₈*f)/(l₄-l₈)；d₄₉=(L₄₉*f)/(l₄-l₉)

d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)；d₅₉=(L₅₉*f)/(l₅-l₉)

d₆₈=(L₆₈*f)/(l₆-l₈)；d₆₉=(L₆₉*f)/(l₆-l₉)；d₇₉=(L₇₉*f)/(l₇-l₉)

3, many camera lenses vision range finding step is obtained

Difference according to 9 each schematic diagrams of camera lens lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, the most finally Range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 8, and j is from 2 to 9, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

Eight, the shooting of more lens stereo visions and disparity computation thereof

1, initialization step

1.1 camera lenses lay mode.Ten camera lenses are with up to 2500 lens stereo vision shootings and disparity computation thereof, and arrangement mode is pressed Following methods obtains:

1) note camera lens number is N, 10≤N≤2500；

2) arrangement maximum number of lines M:M is calculated²≤N<(M+1)², M is an integer determining of corresponding N and 3≤M≤50；

3) selected line number H to be arranged, 1≤H≤M；

3.1) if H=1, the most each camera lens constitutes straight line and equidistant arrangement；

3.2) if H=2, in mode below select one:

The most upper and lower 2 row are parallel, and adjacent nearest every 4 camera lenses constitute square, are similar to " 8 camera lens-7 "；

The most upper and lower 2 row are parallel, and adjacent nearest every 3 camera lenses constitute equilateral triangle, are similar to " 8 camera lens-6 "；

If 3.2.3 N is odd number, then last camera lens 2 camera lenses rightmost with upper and lower 2 row constitute equilateral triangle Shape, is similar to " 9 camera lens-4 ", or direct equidistant cloth is placed on upstream or downstream；

3.3) if H >=3, parallel and equidistant between each row:

3.3.1 according to H, minimum camera lens number I:H × I≤N < H × (I+1) that often row lays is determined；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 4 camera lenses constitute square, likeness in form " 9 camera lens-3 ", or take 3.3.3 mode to lay；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 3 camera lenses constitute equilateral triangle, likeness in form " 9 camera lens-5 "；

3.3.4 K the camera lens (1≤K < H) having more, the most often they cloth are placed on the first row in the middle of line k by row one Extended line on, and rightmost 2 camera lenses of 2 row upper and lower with it constitute equilateral triangle, likeness in form " 9 camera lens-4 "；Or often they equidistant cloth are placed on the first row to the extended line of line k by row one, 4 camera lenses being adjacent constitute square, are similar to " 8 camera lens-8 ".

1.2 initialize all parametric variables

Under the mode that lays of many camera lenses, Si (i is from 1 to N) represents the optical center position of each video camera, represents respectively with Lij (wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j), each video camera for distance between every pair of camera optics center Focal length be f.Make each two video camera on same plane that object P shooting is obtained two dimensional image, according to three-dimensional parallel shooting There is object P parallax on two image surfaces between any two in system vision mode and its computational methods, camera lens i and camera lens j, Being denoted as (li lj), it illustrates P position difference of imaging point in the become image of each two video camera.D represents object P anomaly face Distance, then Si Yu Sj is found range from for dij.

2, all binocular parallax ranging step are obtained

Binocular parallax range finding result of calculation between each two camera lens Si and camera lens Sj is:

d_ij=(L_ij*f)/(l_i-l_j)

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j.

3, many camera lenses vision range finding step is obtained

Laying mode according to each difference of many camera lenses, first obtain each pair of lens pitch from sum: L=∑ Lij, the most finally range finding d is Weighted average is sued for peace:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j.

4, end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The present invention, from the same scenery of multiple viewing point, to obtain the perceptual image under different visual angles, passes through principle of triangulation Calculate the position deviation (i.e. parallax) between image pixel and obtain the three-dimensional information of scenery.At base based on three-dimensional parallel vision mode On plinth, provide the shooting of many lens stereos vision and the computational methods of parallax thereof.Utilize three-dimensional parallel in the environment of multi-lens imaging Many camera lenses are carried out disparity computation, it is thus achieved that the most accurate image disparity, make by the parallax calculation method of vision mode each other Obtain object point depth information in three-dimensional reconstruction the most accurate.

Every kind of many camera lenses laid in mode can rotate around the center of circle according to the ratio that relative distance is constant；Also can be by plane one Straight line is that axle upset lays camera lens.Every kind of many lens stereos vision shooting of the present invention and the computational methods of parallax thereof are by initializing Step, obtain all binocular parallax ranging step, obtain many camera lenses vision range finding step, end step composition.Wherein calculating Have employed average weighted method during final parallax, can preferably eliminate range error.

Every kind of many lens stereos vision shooting of the present invention and the computational methods of parallax thereof by initialization step, obtain all binocular visions Differ from ranging step, obtain many camera lenses vision range finding step, end step composition.Wherein have employed weighting when calculating final parallax Average method, can preferably eliminate range error.

Claims (10)

1. the computational methods of more than lens stereo vision parallax, it is characterised in that including:

One initialized step；

One step obtaining the range finding of all binocular parallaxs；

One step obtaining many camera lenses vision range finding；

One end step.

The computational methods of many lens stereos vision parallax the most as claimed in claim 1, it is characterised in that described many camera lenses be 3 camera lenses, 4 camera lenses, 5 camera lenses, 6 camera lenses, 7 camera lenses, 8 camera lenses, 9 camera lenses and 10～2500 camera lenses.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 3 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 3 camera lenses lay mode

3 camera lenses lay mode and include equilateral triangle, isosceles triangle, arbitrary triangle or the most equidistantly lay；

1.2 initialize all parametric variables

Make each two video camera in three video cameras on same plane that object P shooting is obtained two dimensional image, put down according to solid Row camera system vision mode and its computational methods, there is object P parallax on two image surfaces between any two in three camera lenses, It is denoted as l respectively₁–l₂, l₂–l₃, l₁–l₃, it illustrates P position difference of imaging point in the become image of each two video camera； If d represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₁With S₃Found range from for d₁₃；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)

d₂₃=(L₂₃*f)/(l₂-l₃)

d₁₃=(L₁₃*f)/(l₁-l₃)

3) 3 camera lens vision range finding steps are obtained

3.1 for equilateral triangle, L₁₂=L₂₃=L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.2 for isosceles triangle,The most finally range finding d is:

$d=\frac{2}{5}{d}_{12}+\frac{1}{5}{d}_{23}+\frac{2}{5}{d}_{13}$

3.3 for arbitrary triangle, L₂₃≠L₁₂≠L₁₃, the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃)/3

3.4 for the most equidistantly laying, 2L₂₃=2L₁₂=L₁₃The most finally range finding d is:

$d=\frac{d_{12}+d_{23}}{4}+d_{13}/2$

4) end step

Calculate 3 camera lens vision range finding results according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 4 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 4 camera lenses lay mode

4 camera lenses lay mode and include square；Rectangle；3 camera lens composition equilateral triangles, are positioned on same circle, another Camera lens is in the center of circle；Circle arbitrarily lays；3 camera lens composition equilateral triangles, are positioned on same circle, and another camera lens is waiting The centre of following 2 the camera lens lines of limit triangle；The most equidistantly lay；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄Represent the optical center position of four video cameras, L respectively₁₃, L₁₂, L₂₃, L₁₄, L₂₄, L₃₄Respectively Represent that the distance between every pair of photographic head optical center, the focal length of four video cameras are f, make four video cameras on same plane In each two video camera to object P shooting obtain two dimensional image, according to three-dimensional parallel camera system vision mode and its calculate Method, there is object P parallax on two image surfaces, is denoted as l respectively in four camera lenses between any two₁–l₂, l₂–l₃, l₁–l₃, l₁–l₄, l₂–l₄, l₃–l₄, it illustrates P position difference of imaging point in the become image of each two video camera；D represents object The distance in P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₂With S₄Found range from for d₂₄, S₃With S₄Found range from for d₃₄；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₁₃=(L₁₃*f)/(l₁-L₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₃₄=(L₃₄*f)/(l₃-l₄)

3) 4 camera lens vision range finding steps are obtained

3.1 for square, L₁₂=L₂₃=L₃₄=L₁₄The most finally range finding d is:

$d=\frac{1}{4+2\sqrt{2}}{d}_{12}+\frac{1}{4+2\sqrt{2}}{d}_{23}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{13}+\frac{1}{4+2\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+2\sqrt{2}}{d}_{24}+\frac{1}{4+2\sqrt{2}}{d}_{34}$

3.2 for rectangle,The most finally range finding d is:

$d=\frac{1}{6+2\sqrt{5}}{d}_{12}+\frac{1}{3+\sqrt{5}}{d}_{23}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{13}+\frac{1}{3+\sqrt{5}}{d}_{14}+\frac{\sqrt{5}}{6+2\sqrt{5}}{d}_{24}+\frac{1}{6+2\sqrt{5}}{d}_{34}$

3.3, for 3 camera lenses composition equilateral triangles, are positioned on same circle, another camera lens in the center of circle, L₂₃=L₁₂=L₁₃, The most finally range finding d is:

$d=\frac{1}{3+\sqrt{3}}{d}_{12}+\frac{1}{3+\sqrt{3}}{d}_{23}+\frac{1}{3+\sqrt{3}}{d}_{13}+\frac{1}{3+3\sqrt{3}}{d}_{14}+\frac{1}{3+3\sqrt{3}}{d}_{24}+\frac{1}{3+3\sqrt{3}}{d}_{34}$

3.4 for arbitrarily laying on circle, and camera lens distance between any two is different, and the most finally range finding d is:

D=(d₁₂+d₂₃+d₁₃+d₁₄+d₂₄+d₃₄)/6

3.5, for 3 camera lens composition equilateral triangles, are positioned on same circle, following 2 at equilateral triangle, another camera lens The centre of camera lens line；Or the most equidistantly lay；First obtain each pair of lens pitch from sum: L=L₁₃+L₁₂+L₂₃+L₁₄+L₂₄+L₃₄, the most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 3, and j is from 2 to 4, and i is more than j；

4) end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 5 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 5 camera lenses lay mode

The mode that lays of 5 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another camera lens outer at circle and with just 2 camera lens composition isosceles triangles on square one side；4 camera lens composition squares, are positioned on same circle, another camera lens cloth It is placed on the center of circle；Regular pentagon；The most equidistantly lay；4 camera lens composition squares, are positioned on same circle, another Individual camera lens is in the centre of following 2 the camera lens lines of square；4 lens group rectangularities, another camera lens is in rectangle following 2 The centre of individual camera lens line；Camera lens divides 2 arrangements to put, and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅Represent the optical center position of five video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₁₅, L₁₃, L₁₄, L₂₄, L₂₅, L₃₅Represent that the distance between every pair of camera optics center, the focal length of five video cameras are f respectively；Order is same The each two video camera in five video cameras in one plane obtains two dimensional image to object P shooting, according to the parallel shooting of solid is System vision mode and its computational methods, there is object P parallax on two image surfaces, remember respectively in five camera lenses between any two Make l₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₁–l₅, l₁–l₃, l₁–l₄, l₂–l₄, l₂–l₅, l₃–l₅It illustrates P often The position difference of imaging point in two become images of video camera；D represents the distance in object P anomaly face, then S₁With S₂Found range From for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₁ With S₅Found range from for d₁₅, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₂With S₄Found range from For d₂₄, S₂With S₅Found range from for d₂₅, S₃With S₅Found range from for d₃₅；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₃₅=(L₃₅*f)/(l₃-l₅)

3) 5 camera lens vision range finding steps are obtained

3.1, for 4 camera lenses composition square, are positioned on same circle, another camera lens outer at circle and with 2 of square Camera lens composition isosceles triangle, L₁₂=L₂₃=L₃₄=L₁₄,The most finally range finding d is: $\begin{array}{c}d=\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{12}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{23}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{34}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{45}+\frac{\sqrt{2}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{15}\\ +\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{13}+\frac{1}{4+3\sqrt{2}+\sqrt{10}}{d}_{14}+\frac{\sqrt{2}}{4+3\sqrt{2}+\sqrt{10}}{d}_{24}+\frac{\sqrt{2}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{25}+\frac{\sqrt{10}}{8+6\sqrt{2}+2\sqrt{10}}{d}_{35}\end{array}$

3.2, for 4 camera lens composition squares, are positioned on same circle, and another camera lens cloth is placed on the center of circle, L₁₂=L₃₄=L₁₄=L₂₃The most finally range finding d is:

$\begin{array}{c}d=\frac{1}{4+4\sqrt{2}}{d}_{12}+\frac{1}{4+4\sqrt{2}}{d}_{23}+\frac{1}{4+4\sqrt{2}}{d}_{34}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{45}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{15}\\ +\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{13}+\frac{1}{4+4\sqrt{2}}{d}_{14}+\frac{\sqrt{2}}{4+4\sqrt{2}}{d}_{24}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{25}+\frac{\sqrt{2}}{8+8\sqrt{2}}{d}_{35}\end{array}$

3.3 for regular pentagon, L₁₂=L₂₃=L₃₄=L₄₅=L₁₅, the most finally range finding d is:

$\begin{array}{c}d=\frac{1}{5+5\mathrm{\&alpha;}}{d}_{12}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{23}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{34}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{45}+\frac{1}{5+5\mathrm{\&alpha;}}{d}_{15}\\ +\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{13}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{14}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{24}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{25}+\frac{\mathrm{\&alpha;}}{5+5\mathrm{\&alpha;}}{d}_{35}\end{array}$

Wherein, α represents d₁₃、d₁₄、d₂₄、d₂₅、d₃₅Binocular range measurement at the ratio of all two camera lens binocular range measurement summations Example, takes any value (it is said that in general, value is more than or equal to 1) between 0.6 to 2.4, desirable incremental step a length of 0.1；

3.4 for the most equidistantly laying；Or 4 camera lens composition squares, it is positioned on same circle, another camera lens Centre at following 2 the camera lens lines of square；Or 4 lens group rectangularities, following 2 in rectangle, another camera lens The centre of camera lens line；Or camera lens divides 2 arrangements to put, adjacent camera lens all forms equilateral triangle；First obtain each pair of lens pitch from Sum: L=L₁₂+L₂₃+L₃₄+L₄₅+L₁₅+L₁₃+L₁₄+L₂₄+L₂₅+L₃₅, the most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 4, and j is from 2 to 5, and i is more than j；

4) end step

Calculate 5 camera lens vision range finding results according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 6 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 6 camera lenses lay mode

The mode that lays of 6 camera lenses includes: 4 camera lens composition squares, is positioned on same circle, and another 2 camera lenses are outer and difference at circle 2 camera lenses adjacent with square both sides and the center of circle respectively form 2 squares；4 camera lens composition squares, are positioned at same circle On, a camera lens cloth is placed on the center of circle, and outer at circle and with square below 2 camera lenses of another camera lens form isosceles Triangle；5 camera lens composition regular pentagons, are positioned on same circle, and another 1 camera lens cloth is placed on the center of circle；Equilateral hexagon, position On same circle；The most equidistantly lay；4 camera lens composition squares, are positioned on same circle, and another 2 camera lenses are each The centre of following 2 camera lens lines on square；6 camera lens rectangles；3 camera lens one big equilateral triangles in outside of composition, Another 3 cloth are placed on the centre of big equilateral triangle each edge, the little equilateral triangle that composition stands upside down；2 arrangements are divided to put, adjacent Camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆Represent the optical center position of six video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₁₆, L₁₃, L₁₄, L₁₅, L₂₄, L₂₅, L₂₆, L₃₅, L₃₆, L₄₆Represent the distance between six camera optics centers respectively, The focal length of six video cameras is f；Make each two video camera in six video cameras on same plane that object P is shot acquisition two Dimension image, according to three-dimensional parallel camera system vision mode and its computational methods, there is object P and exist in six camera lenses between any two Parallax on two image surfaces, is denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₁–l₆, l₁–l₃, l₁–l₄, l₁–l₅, l₂–l₄, l₂–l₅, l₂–l₆, l₃–l₅, l₃–l₆, l₄–l₆It illustrates P and becomes in the become image of each two video camera The position difference of picture point；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range From for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₁ With S₆Found range from for d₁₆, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from For d₁₅, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₄With S₆Found range from for d₄₆；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l3)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₁₆=(L₁₆*f)/(l₁-l₆)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅, f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₄₆=(L₄₆*f)/(l₄-l₆)

3) 6 camera lens vision range finding steps are obtained

Difference according to 6 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄+ L₄₅+L₅₆+L₁₆+L₁₃+L₁₄+L₁₅+L₂₄+L₂₅+L₂₆+L₃₅+L₃₆+L₄₆, the most finally range finding d is weighted average summation:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 5, and j is from 2 to 6, and i is more than j；

4) end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 7 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 7 camera lenses lay mode

The summit of 7 camera lenses lays mode and includes: 4 camera lens composition squares, is positioned on same circle, and another 3 camera lenses are outer also at circle It is adjacent 2 camera lenses on square limit and the center of circle respectively forms 3 squares；4 camera lens composition squares, are positioned at same On circle, a camera lens cloth is placed on the center of circle, another 2 outer at circle and be adjacent 2 camera lenses on square limit and the center of circle respectively forms 2 squares；4 camera lenses composition square, is positioned on same circle, and a camera lens cloth is placed on the center of circle, another 2 outer at circle and with 2 camera lenses on its adjacent square limit respectively form 2 isosceles triangles；6 camera lens composition equilateral hexagon, are positioned at same On circle, a camera lens cloth is placed on the center of circle；Form equilateral heptagon, be positioned on same circle；The most equidistantly lay；4 Individual camera lens composition square, is positioned on same circle, another 2 camera lenses respectively centre of following 2 camera lens lines on square, 1 Individual camera lens is on the right and 2 camera lenses being adjacent form equilateral or isosceles triangles；4 camera lens composition squares, are positioned at same On one circle, another 2 camera lenses respectively centre of following 2 camera lens lines on square, another 1 camera lens is on top and and the upper left corner Equilateral or isosceles triangle is formed with the 2 of the upper right corner camera lenses；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle； Divide 2 arrangements to put and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆, S₇Represent the optical center position of seven video cameras, L respectively₁₂, L₂₃, L₃₄, L₄₅, L₅₆, L₆₇, L₁₇, L₁₃, L₁₄, L₁₅, L₁₆, L₂₄, L₂₅, L₂₆, L₂₇, L₃₅, L₃₆, L₃₇, L₄₆, L₄₇, L₅₇Represent respectively Distance between every pair of camera optics center, the focal length of seven video cameras is f, makes in seven video cameras on same plane Each two video camera obtains two dimensional image to object P shooting, according to three-dimensional parallel camera system vision mode and its computational methods, There is object P parallax on two image surfaces in seven camera lenses, is denoted as l between any two respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₁–l₇, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₃–l₅, l₃–l₆, l₃–l₇, l₄–l₆, l₄–l₇, l₅–l₇, it illustrates P position difference of imaging point in the become image of each two video camera；d Represent the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄ Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from for d₆₇, S₁With S₇Found range from for d₁₇, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range From for d₁₅, S₁With S₆Found range from for d₁₆, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂ With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from For d₃₆, S₃With S₇Found range from for d₃₇, S₄With S₆Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₅With S₇Found range from for d₅₇；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₁₇=(L₁₇*f)/(l₁-l₇)；d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)

d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*f)/(l₁-l₆)；d₂₄=(L₂₄*f)/(l₂-l₄)

d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)；d₅₇=(L₅₇*f)/(l₅-l₇)

3) 7 camera lens vision range finding steps are obtained

Difference according to 7 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=L₁₂+L₂₃+L₃₄ +L₄₅+L₅₆+L₆₇+L₁₇+L₁₃+L₁₄+L₁₅+L₁₆+L₂₄+L₂₅+L₂₆+L₂₇+L₃₅+L₃₆+L₃₇+L₄₆+L₄₇+L₅₇, the most finally find range d Sue for peace for weighted average:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 6, and j is from 2 to 7, and i is more than j；

4) end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 8 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 8 camera lenses lay mode

The mode that lays of 8 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another 3 camera lenses outer at circle and and its 2 camera lenses and the center of circle on adjacent square limit respectively form 3 squares, and another 1 camera lens lays in Far Left symmetry；4 Camera lens composition square, is positioned on same circle, and another 3 camera lenses are outer at circle and are adjacent 2 camera lenses on square limit and circle The heart respectively forms 3 squares, and another 1 camera lens cloth is placed on the center of circle；7 camera lenses form equilateral heptagon, are positioned on same circle, Another camera lens cloth is placed on the center of circle；8 camera lenses form equilateral octagon, are positioned on same circle；The most equidistantly lay； Divide 2 arrangements to put and adjacent camera lens all forms equilateral triangle；Divide 2 arrangements to put and adjacent camera lens all forms square；Divide 3 Arrangement is put and adjacent camera lens all forms square；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈Represent the optical center position of eight video cameras respectively, use L_ijRepresent respectively (wherein i is from 1 to 7, and j is from 2 to 8, and i is more than j), Jiao of eight video cameras for distance between every pair of camera optics center Away from being f, each two video camera in eight video cameras on same plane is made object P shooting to be obtained two dimensional image, according to vertical Body parallel camera system vision mode and its computational methods, there is object P on two image surfaces in eight camera lenses between any two Parallax, is denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇–l₈, l₁–l₈, l₁–l₃, l₁–l₄, l₁– l₅, l₁–l₆, l₁–l₇, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₄–l₆, l₄–l₇, l₄–l₈, l₅–l₇, l₅–l₈, l₆–l₈, it illustrates P alternate position spike of imaging point in the become image of each two video camera Different；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃ With S₄Found range from for d₃₄, S₄With S₅Found range from for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from For d₆₇, S₇With S₈Found range from for d₇₈, S₁With S₈Found range from for d₁₈, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from for d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₁With S₇Found range from for d₁₇, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇ Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₃With S₅Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₄With S₆Found range from for d₄₆, S₄With S₇Found range From for d₄₇, S₄With S₈Found range from for d₄₈, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₆ With S₈Found range from for d₆₈；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₁₈=(L₁₈*f)/(l₁-l₈)；d₁₃=(L₁₃*f)/(l₁-l₃)

d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)；d₁₆=(L₁₆*f)/(l₁-l₆)

d₁₇=(L₁₇*f)/(l₁-l₇)；d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)

d₂₆=(L₂₆*f)/(l₂-l₆)；d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)

d₃₅=(L₃₅*f)/(l₃-l₅)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₄₆=(L₄₆*f)/(l₄-l₆)；d₄₇=(L₄₇*f)/(l₄-l₇)

d₄₈=(L₄₈*f)/(l₄-l₈)；d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)

d₆₈=(L₆₈*f)/(l₆-l₈)

3) 8 camera lens vision range finding steps are obtained

Difference according to 8 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, and the most finally range finding d is for adding Weight average is sued for peace:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to 7, and j is from 2 to 8, and i is more than j；

4) end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 9 lens stereo visions Parallax calculation method comprises the following steps:

1) initialization step

1.1 9 camera lenses lay mode

The mode that lays of 9 camera lenses includes: 4 camera lenses composition square, is positioned on same circle, another 3 camera lenses outer at circle and and its 2 camera lenses and the center of circle on adjacent square limit respectively form 3 squares, and 1 camera lens lays in Far Left symmetry, 1 camera lens Cloth is placed on the center of circle；8 camera lenses form equilateral octagon, and 1 camera lens cloth is placed on the center of circle；3 rows are divided equidistantly to lay, adjacent mirror Head all forms square；8 camera lenses divide 2 rows equidistantly to lay, and all form square, another 1 camera lens on the right and with its phase 2 adjacent camera lenses form equilateral or isosceles triangle；Divide 3 arrangements to put and adjacent camera lens all forms equilateral triangle；Divide 2 rows Lay and adjacent camera lens all forms equilateral triangle；The most equidistantly lay；

1.2 initialize all parametric variables

S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈, S₉Represent the optical center position of nine video cameras respectively, with Lij respectively (wherein i is from 1 to 8, and j is from 2 to 9, and i is more than j), nine video cameras to represent the distance between every pair of camera optics center Focal length be f, make each two video camera in nine video cameras on same plane that object P shooting is obtained two dimensional image, root According to three-dimensional parallel camera system vision mode and its computational methods, there is object P between any two at two image surfaces in nine camera lenses On parallax, be denoted as l respectively₁–l₂, l₂–l₃, l₃–l₄, l₄–l₅, l₅–l₆, l₆–l₇, l₇–l₈, l₈–l₉, l₁–l₉, l₁–l₃, l₁–l₄, l₁–l₅, l₁–l₆, l₁–l₇, l₁–l₈, l₂–l₄, l₂–l₅, l₂–l₆, l₂–l₇, l₂–l₈, l₂–l₉, l₃–l₅, l₃–l₆, l₃–l₇, l₃–l₈, l₃–l₉, l₄–l₆, l₄–l₇, l₄–l₈, l₄–l₉, l₅–l₇, l₅–l₈, l₅–l₉, l₆–l₈, l₆–l₉, l₇–l₉ It illustrates P position difference of imaging point in the become image of each two video camera；D represents the distance in object P anomaly face, then S₁With S₂Found range from for d₁₂, S₂With S₃Found range from for d₂₃, S₃With S₄Found range from for d₃₄, S₄With S₅Found range From for d₄₅, S₅With S₆Found range from for d₅₆, S₆With S₇Found range from for d₆₇, S₇With S₈Found range from for d₇₈, S₈ With S₉Found range from for d₈₉, S₁With S₉Found range from for d₁₉, S₁With S₃Found range from for d₁₃, S₁With S₄Found range from For d₁₄, S₁With S₅Found range from for d₁₅, S₁With S₆Found range from for d₁₆, S₁With S₇Found range from for d₁₇, S₁With S₈Found range from for d₁₈, S₂With S₄Found range from for d₂₄, S₂With S₅Found range from for d₂₅, S₂With S₆Found range from for d₂₆, S₂With S₇Found range from for d₂₇, S₂With S₈Found range from for d₂₈, S₂With S₉Found range from for d₂₉, S₃With S₅ Found range from for d₃₅, S₃With S₆Found range from for d₃₆, S₃With S₇Found range from for d₃₇, S₃With S₈Found range from for d₃₈, S₃With S₉Found range from for d₃₉, S₄With S₆Found range from for d₄₆, S₄With S₇Found range from for d₄₇, S₄With S₈Found range From for d₄₈, S₄With S₉Found range from for d₄₉, S₅With S₇Found range from for d₅₇, S₅With S₈Found range from for d₅₈, S₅ With S₉Found range from for d₅₉, S₆With S₈Found range from for d₆₈, S₆With S₉Found range from for d₆₉, S₇With S₉Found range from For d₇₉；

2) all binocular parallax ranging step are obtained

The binocular parallax range finding result of calculation of each two camera lens is respectively as follows:

d₁₂=(L₁₂*f)/(l₁-l₂)；d₂₃=(L₂₃*f)/(l₂-l₃)；d₃₄=(L₃₄*f)/(l₃-l₄)

d₄₅=(L₄₅*f)/(l₄-l₅)；d₅₆=(L₅₆*f)/(l₅-l₆)；d₆₇=(L₆₇*f)/(l₆-l₇)

d₇₈=(L₇₈*f)/(l₇-l₈)；d₈₉=(L₈₉*f)/(l₈-l₉)；d₁₉=(L₁₉*f)/(l₁-l₉)

d₁₃=(L₁₃*f)/(l₁-l₃)；d₁₄=(L₁₄*f)/(l₁-l₄)；d₁₅=(L₁₅*f)/(l₁-l₅)

d₁₆=(L₁₆*f)/(l₁-l₆)；d₁₇=(L₁₇*f)/(l₁-l₇)；d₁₈=(L₁₈*f)/(l₁-l₈)

d₂₄=(L₂₄*f)/(l₂-l₄)；d₂₅=(L₂₅*f)/(l₂-l₅)；d₂₆=(L₂₆*f)/(l₂-l₆)

d₂₇=(L₂₇*f)/(l₂-l₇)；d₂₈=(L₂₈*f)/(l₂-l₈)；d₂₉=(L₂₉*f)/(l₂-l₉)

d₃₅=(L₃₅*f)/(l₃-l5)；d₃₆=(L₃₆*f)/(l₃-l₆)；d₃₇=(L₃₇*f)/(l₃-l₇)

d₃₈=(L₃₈*f)/(l₃-l₈)；d₃₉=(L₃₉*f)/(l₃-l₉)；d₄₆=(L₄₆*f)/(l₄-l₆)

d₄₇=(L₄₇*f)/(l₄-l₇)；d₄₈=(L₄₈*f)/(l₄-l₈)；d₄₉=(L₄₉*f)/(l₄-l₉)

d₅₇=(L₅₇*f)/(l₅-l₇)；d₅₈=(L₅₈*f)/(l₅-l₈)；d₅₉=(L₅₉*f)/(l₅-l₉)

d₆₈=(L₆₈*f)/(l₆-l₈)；d₆₉=(L₆₉*f)/(l₆-l₉)；d₇₉=(L₇₉*f)/(l₇-l₉)

3) 9 camera lens vision range finding steps are obtained

Difference according to 9 camera lenses lays mode, first obtains each pair of lens pitch from sum: L=∑ Lij, and the most finally range finding d is for adding Weight average is sued for peace:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1～8, and j is from 2～9, and i is more than j；

4) end step

Calculate many camera lenses vision range finding result according to as above formula, result is exported.

The computational methods of many lens stereos vision parallax the most as claimed in claim 2, it is characterised in that described 10～2500 camera lenses Stereoscopic vision parallax calculation method comprises the following steps:

1) initialization step

1.1 camera lenses lay mode

The arrangement mode of 10～2500 camera lenses obtains by the following method:

1) note camera lens number is N, 10≤N≤2500；

2) arrangement maximum number of lines M:M is calculated²≤N<(M+1)², M is an integer determining of corresponding N and 3≤M≤50；

3) selected line number H to be arranged, 1≤H≤M；

3.1) if H=1, the most each camera lens constitutes straight line and equidistant arrangement；

3.2) if H=2, in mode below select one:

The most upper and lower 2 row are parallel, and adjacent nearest every 4 camera lenses constitute square；

The most upper and lower 2 row are parallel, and adjacent nearest every 3 camera lenses constitute equilateral triangle；

If 3.2.3 N is odd number, then last camera lens 2 camera lenses rightmost with upper and lower 2 row constitute equilateral triangle Shape, or direct equidistant cloth is placed on upstream or downstream；

3.3) if H >=3, parallel and equidistant between each row:

3.3.1 according to H, minimum camera lens number I:H × I≤N < H × (I+1) that often row lays is determined；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 4 camera lenses constitute square, or take 3.3.3 Mode lays；

In the row that the most each upper and lower 2 row are adjacent, adjacent nearest every 3 camera lenses constitute equilateral triangle；

3.3.4 K the camera lens (1≤K < H) having more, the most often they cloth are placed on the first row to middle the prolonging of line k by row one On long line, and rightmost 2 camera lenses of 2 row upper and lower with it constitute equilateral triangle；Or often row one is by they equidistant cloth Being placed on the first row to the extended line of line k, 4 camera lenses being adjacent constitute square；

1.2 initialize all parametric variables

Under the mode that lays of 10～2500 camera lenses, Si (i is from 1 to N) represents the optical center position of each video camera, divides with Lij Do not represent distance between every pair of camera optics center (wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j), each The focal length of video camera is f, makes each two video camera on same plane that object P shooting is obtained two dimensional image, puts down according to solid There is object P between any two on two image surfaces in row camera system vision mode and its computational methods, camera lens i and camera lens j Parallax, be denoted as (li lj), it illustrates P position difference of imaging point in the become image of each two video camera；D represents object The distance in P anomaly face, then Si Yu Sj is found range from for dij；

2) all binocular parallax ranging step are obtained

Binocular parallax range finding result of calculation between each two camera lens Si and camera lens Sj is:

d_ij=(L_ij*f)/(l_i-l_j)

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j；

3) 10～2500 camera lens vision range finding steps are obtained

Lay mode according to 10～2500 each differences of camera lens, first obtain each pair of lens pitch from sum: L=∑ Lij, the most finally survey It is weighted average summation away from d:

$d=\mathrm{\&Sigma;}\frac{L_{ij}}{L}{d}_{ij}$

Wherein i is from 1 to N-1, and j is from 2 to N, and i is more than j；

4) end step

Calculate 10～2500 camera lens vision range finding results according to as above formula, result is exported.