CN113256812B - Characteristic light ray sampling optimization method and processing device for light field camera and storage medium - Google Patents
Characteristic light ray sampling optimization method and processing device for light field camera and storage medium Download PDFInfo
- Publication number
- CN113256812B CN113256812B CN202110376725.0A CN202110376725A CN113256812B CN 113256812 B CN113256812 B CN 113256812B CN 202110376725 A CN202110376725 A CN 202110376725A CN 113256812 B CN113256812 B CN 113256812B
- Authority
- CN
- China
- Prior art keywords
- light
- characteristic
- rays
- flame
- ray
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
- G06T17/205—Re-meshing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
Abstract
The invention discloses a characteristic light ray sampling optimization method, a processing device and a storage medium of a light field camera, wherein the characteristic light ray sampling optimization method comprises the following steps: carrying out three-dimensional gridding on the flame; tracking light rays from an image detector to flames according to camera parameters, and recording a flame grid number through which each light ray passes, and a zenith angle and a circumferential angle of each light ray; retaining the effective light rays passing through the flame grid; carrying out ray clustering and circumferential angle clustering on the effective rays; traversing each cluster in the circumferential angle clusters, and each cluster in the light clusters, determining the final characteristic light. The invention does not depend on the physical parameters of flame and the type of the light field camera, and has simple operation and easy realization. In the process of reconstructing the flame temperature field, the temperature field can be rapidly and accurately reconstructed only by utilizing the characteristic light, the problem of huge time cost caused by huge and redundant flame radiation information in the process of solving the temperature field is solved, and the requirement on the memory of a computer is reduced.
Description
Technical Field
The invention relates to a characteristic light ray sampling optimization method of a light field camera, and belongs to the technical field of flame temperature measurement.
Background
Flame combustion temperature is an important indicator that is closely related to the combustion process. The measurement of the flame three-dimensional temperature field is helpful for revealing the nature of the combustion phenomenon and the law of the combustion process, and has great significance for controlling the generation and the emission of combustion pollutants and the design and the optimized operation of a combustion system. The flame temperature field non-contact measurement technology based on radiation imaging utilizes the self radiation image of flame to obtain the temperature information of the flame, has the advantages of non-invasive, high measurement precision, continuous real-time measurement and the like, and is researched and used by more and more researchers recently.
The light field camera is as flame radiation imaging collection system, installs the microlens array additional in the middle of main lens and image detector, compares in traditional camera, not only can record flame radiation's intensity information with higher accuracy under single exposure, can also distinguish the direction of radiation light, combines the inversion algorithm, if the reconstruction of the three-dimensional temperature field of flame just can be realized to nonnegative least square algorithm.
The light field camera detects a large amount of light, but the light field camera separates the radiation intensity information in different directions focused on an image point on the microlens by the main lens by using the microlens array and records the separated radiation intensity information on the pixels of the CCD detector covered behind the microlens, so that the angle variation range of the direction of the separated detected light is small (for example, the cone angle of each pixel of the image detector of the Raytrix camera corresponding to the light beam is less than 0.015 °), and the redundancy of the light is high. When the flame three-dimensional temperature field is reconstructed by using the huge and redundant flame radiation information, the time cost brought by the complex matrix operation and the requirement of large-scale matrix storage on the memory of a computer cause that the time resolution of the reconstruction process is lower and the resource consumption is larger, and particularly when the flame temperature field, the absorption coefficient distribution and the like are jointly reconstructed, the reconstruction time is longer and the required computing resource is larger. It is therefore necessary to optimize the light field sampling, reducing the redundancy of the light field sampling.
Disclosure of Invention
The present invention provides a method, a processing device and a storage medium for optimizing characteristic light ray sampling of a light field camera, aiming at the deficiencies of the prior art.
In order to solve the technical problems, the invention adopts the technical scheme that:
a method for optimizing the sampling of characteristic light rays of a light field camera is characterized by comprising the following steps:
step one, carrying out three-dimensional grid division on flame, and numbering the divided flame grids;
secondly, tracking one light ray by each pixel according to camera parameters, tracking the light ray to flame from an image detector, and recording the number of flame grids penetrated by each light ray and the zenith angle and the circumferential angle of each light ray;
deleting ineffective light rays which do not pass through the flame grid, and keeping the effective light rays which pass through the flame grid; the effective light rays are processed in the fourth step to the ninth step;
step four, light clustering: classifying the effective light rays passing through the same flame grid into one class;
fifth, circumferential angle clustering: aiming at each type in the fourth step, performing circumference angle clustering according to the distribution of circumference angles of each type of light, classifying the light rays with the same circumference angle into a cluster, and obtaining zenith angle distribution of each cluster of light rays which pass through the same flame grid and have the same circumference angle;
step six, selecting K characteristic zenith angles in the zenith angles corresponding to the cluster of light rays in the step five;
step seven, selecting K rays corresponding to the characteristic zenith angles in the step six as characteristic rays of the cluster of rays;
and step eight, traversing each cluster in the circumferential angle clusters and each cluster in the light clusters to determine the final characteristic light.
The final characteristic ray total determined is:
wherein P is the total number of characteristic rays, M is the number of classes of effective ray clusters, and N m Indicating the number of clusters in the mth class of rays.
K takes a value of 2-5.
K takes a value of 3.
In the sixth step, the 3 characteristic zenith angles are respectively the largest and smallest zenith angles and the zenith angle with the smallest difference value with the average value.
The camera parameters are obtained by calibration.
Establishing a right-hand Cartesian coordinate system in the flame center, wherein the z axis is the height direction of the flame, the x axis points to the camera, and the y axis is the horizontal direction; the included angle between the ray and the z axis is defined as a zenith angle, and the included angle between the projection line of the ray on the x-y plane and the x axis is defined as a circumferential angle.
A light field camera's characteristic light sampling optimizing apparatus, characterized in that includes:
the light ray tracking module is used for tracking one light ray for each pixel according to camera parameters, tracking the light ray to flame from the image detector, and recording the number of a flame grid through which each light ray passes, and the zenith angle and the circumferential angle of the light ray;
the effective light selecting module selects light rays penetrating through the flame grid as effective light rays;
the light clustering module is used for classifying the effective light rays passing through the same flame grid in the effective light rays; clustering the light rays with the same circumferential angle in each classified effective light ray;
and the light selecting module selects the light rays of each divided effective light cluster corresponding to K characteristic zenith angles as final characteristic light rays.
A processing apparatus, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method for feature ray sampling optimization.
A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, implements the method for optimizing feature ray sampling.
The calibration of the internal and external parameters of the light field camera is carried out according to the method disclosed by the Chinese patent invention with the publication number of CN105488810B and the name of the invention of a method for calibrating the internal and external parameters of the focusing light field camera.
In the fifth step, because the change rate of the height direction of the flame is larger than that of the radial direction, the change rate of the zenith angle of each type of light in the fourth step is larger than that of the circumferential angle, and therefore the fifth step clusters the circumferential angles.
In the sixth step, in order to ensure that the angle information of the optimized light is not redundant and not lost, the characteristic zenith angles selected for each cluster of light are respectively the largest zenith angle, the smallest zenith angle and the zenith angle with the smallest difference value with the average value.
Has the advantages that:
the invention provides a characteristic light ray sampling optimization method and a processing device aiming at the characteristics of huge sampling light ray quantity and redundancy of a light field camera. By tracing the sampling light, performing light clustering and circumferential angle clustering according to the number of the flame grids penetrated by the light and the angle information of the light, screening out redundant sampling light, and quickly and accurately reconstructing the temperature field by using the characteristic light in the reconstruction process of the flame temperature field, the problem of huge time cost caused by huge and redundant flame radiation information in the solving process of the temperature field is solved, and the requirement on the memory of a computer is reduced.
Drawings
FIG. 1 is a schematic view of flame meshing and mesh numbering;
FIG. 2 is a flow chart of a light field camera characteristic light field sampling optimization method;
FIG. 3 is a schematic diagram of light field camera ray tracing;
FIG. 4 is a schematic diagram of coordinate system establishment, circumferential angle, zenith angle;
FIG. 5 is an effective ray distribution diagram;
FIG. 6 is a characteristic ray profile;
fig. 7 is an error map of the temperature field reconstruction.
Detailed Description
The invention is further illustrated by the following specific examples in conjunction with the accompanying drawings. It is to be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention, which is to be given the full breadth of the appended claims and any and all equivalent modifications within the scope of the present invention as defined by the following claims.
A method for optimizing characteristic ray sampling of a light field camera comprises the following steps:
step one, setting light field camera parameters:
the focal lengths of the main lens and the micro-lens are respectively 50mm and 0.6mm, the distance from the flame center to the main lens is 505mm, the distance from the main lens to the micro-lens is 53.1mm, the distance from the lens to the image detector is 0.48mm, each micro-lens covers 12 pixels, the size of each pixel is 0.008mm multiplied by 0.008mm, and the resolution of the image detector is 720 (rows) multiplied by 720 (columns).
Step two, setting flame parameters:
the flame is arranged into a cylindrical flame, and the division and numbering schematic diagram of the flame grid is shown in figure 1. The flame base radius is R =0.008m and the height is Z =0.03m. Divide the flame into
(circumferential direction) × Nr (radial direction) × Nz (axial direction) =10 × 8 × 6=480 meshes. The flame internal temperature distribution obeys equation (1).
Where T is the flame temperature and is given in K, x, y and z are the x, y and z directions of the coordinate system, respectively.
The flow chart of the characteristic light ray sampling optimization method of the light field camera is shown in fig. 2, and the step three to the step nine complete the characteristic light ray sampling optimization of the light field camera according to the flow chart shown in fig. 2.
Step three, ray tracing:
according to the ray tracing diagram shown in fig. 3, each pixel traces a ray, the ray is traced from the image detector to the flame, and the number of the flame grid traversed by each ray, and the zenith angle and the circumferential angle of the ray are recorded, and the establishment of the coordinate system and the definition of the zenith angle and the circumferential angle are shown in fig. 4;
step four, deleting the invalid light rays which do not pass through the flame grid, and keeping the valid light rays which pass through the flame grid, as shown in fig. 5, the light rays corresponding to the brightened pixels are valid light rays, and the rest are invalid light rays, in this example, the total number of the statistical light rays is 518400, the valid light rays are 28867, and the rest are invalid light rays;
step five, light clustering: the effective light rays passing through the same flame grid are classified into M types. In this example 28867 significant rays are counted through the number of flame grids, there is 881 class of rays through the different flame grids, then M =881;
sixthly, clustering circumferential angles: for each type in the fifth step, performing circumference angle clustering according to the distribution of circumference angles of each type of light, and classifying the light rays with equal circumference angles into a cluster, wherein each type of light rays has N clusters, and N is total m Representing the number of light ray clusters in the mth type light rays to obtain the zenith angle distribution of each light ray cluster which passes through the same flame grid and has the same circumferential angle;
step seven, selecting K characteristic zenith angles in the zenith angles corresponding to the cluster of light rays in the step six;
step eight, selecting K rays corresponding to the K characteristic zenith angles in the step seven as characteristic rays of the cluster of rays;
step nine, traversing each cluster in the circumferential angle clusters and each class in the light clusters, and determining the final total number P of the characteristic light rays:
in one embodiment, K is 3,3 characteristic zenith angles, which are the largest, the smallest and the three zenith angles having the smallest difference from the average value.
When the value of K is 3, traversing each cluster in the circumferential angle clusters and each cluster in the light clusters, and determining the final characteristic light total number P:
this example yields a final characteristic ray count of 5070 according to equation (3), and the distribution of characteristic rays is shown in fig. 6, where the rays corresponding to the highlighted pixels are characteristic rays.
For fully comparing the effectiveness of the present invention in reconstructing the temperature field, the cases of K =2, 4 and 5 were selected at the same time, and compared with the case of K =3, when K was selected to be 2, 3,4 and 5, respectively, the total number of corresponding characteristic rays was 3380, 5070, 6760 and 8450, respectively.
And establishing a radiation transfer equation according to the radiation information of the selected characteristic light, and solving the radiation transfer equation by using an lsqnanneg function of Matlab. And calculating the relative error delta T of the reconstruction temperature of each grid according to the formulas (4) and (5) i And average relative error Δ T of all grid reconstruction temperatures mean ：
In the formula,. DELTA.T i Reconstructing the relative error, T, of the flame temperature for the ith mesh rst,i Flame temperature value, T, reconstructed for ith grid set,i And setting the flame temperature value for the ith grid, wherein Q is the total number of the flame grids.
When K is respectively selected to be 2, 3,4 and 5, the finally selected characteristic light is used for reconstructing the temperature distribution of the formula (1), and the required time is 31s,60s,95s and 134s respectively. The maximum and average relative error of the reconstructed temperature is shown in fig. 7. It can be seen that when K =2, the reconstruction accuracy is lowest at this time, the maximum relative error is 0.710, and the average relative error is 0.081. As K increases, the reconstruction time increases and the reconstruction error decreases. The maximum and average relative errors of the reconstruction are almost identical when K =3,4 and 5, but the reconstruction time only needs 60s when K = 3. In summary, the three zenith angles with the largest and smallest values and the smallest difference value with the average value among the zenith angles corresponding to the cluster of light rays are selected as the optimal scheme of the characteristic zenith angle.
Claims (9)
1. A method for optimizing the sampling of characteristic light rays of a light field camera is characterized by comprising the following steps:
step one, performing three-dimensional grid division on flame, and numbering the divided flame grids;
secondly, tracking one light ray by each pixel according to camera parameters, tracking the light ray to flame from an image detector, and recording the number of flame grids penetrated by each light ray and the zenith angle and the circumferential angle of each light ray;
deleting ineffective light rays which do not pass through the flame grid, and keeping the effective light rays which pass through the flame grid; the effective light rays are processed in the fourth step to the ninth step;
step four, light clustering: classifying the effective light rays passing through the same flame grid;
fifth, circumferential angle clustering: aiming at each type in the fourth step, performing circumference angle clustering according to the distribution of circumference angles of each type of light, classifying the light rays with the same circumference angle into a cluster, and obtaining zenith angle distribution of each cluster of light rays which pass through the same flame grid and have the same circumference angle;
step six, selecting K characteristic zenith angles in the zenith angles corresponding to the cluster of light rays in the step five;
step seven, selecting K rays corresponding to the characteristic zenith angles in the step six as characteristic rays of the cluster of rays;
and step eight, traversing each cluster in the circumferential angle clusters and each class in the light clusters to determine the final characteristic light.
2. The method of claim 1, wherein the determined final total number of characteristic rays is:
wherein P is the total number of characteristic rays, M is the number of classes of effective ray clusters, and N m Indicating the number of clusters in the mth class of rays.
3. The method for optimizing the sampling of characteristic rays of a light field camera according to claim 1, wherein K takes a value of 2 to 5.
4. The method for optimizing the sampling of characteristic rays of a light field camera according to claim 3, wherein K takes a value of 3.
5. The method for optimizing the sampling of characteristic rays of a light field camera according to claim 4, wherein in the sixth step, the 3 characteristic zenith angles are respectively the maximum and minimum zenith angles and the zenith angle with the minimum difference value from the average value.
6. The method for optimizing the sampling of characteristic rays of a light field camera according to claim 1 wherein the camera parameters are obtained by calibration.
7. The method for optimizing the sampling of the characteristic rays of the light field camera according to claim 1, characterized in that a right-handed cartesian coordinate system is established at the center of the flame, the z-axis is the height direction of the flame, the x-axis points to the camera, and the y-axis is the horizontal direction; the included angle of the ray and the z axis is defined as a zenith angle, and the included angle of the projection line of the ray on the x-y plane and the x axis is defined as a circumferential angle.
8. A processing apparatus, comprising: at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the characteristic ray sampling optimization method as claimed in any one of claims 1 to 7.
9. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method for optimizing a characteristic ray sampling according to one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110376725.0A CN113256812B (en) | 2021-04-07 | 2021-04-07 | Characteristic light ray sampling optimization method and processing device for light field camera and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110376725.0A CN113256812B (en) | 2021-04-07 | 2021-04-07 | Characteristic light ray sampling optimization method and processing device for light field camera and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113256812A CN113256812A (en) | 2021-08-13 |
| CN113256812B true CN113256812B (en) | 2022-11-04 |
Family
ID=77220484
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110376725.0A Active CN113256812B (en) | 2021-04-07 | 2021-04-07 | Characteristic light ray sampling optimization method and processing device for light field camera and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113256812B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608738A (en) * | 2016-03-04 | 2016-05-25 | 华北电力大学（保定） | Light field camera-based flame three-dimensional photometric field reconstruction method |
| CN108389169A (en) * | 2018-03-07 | 2018-08-10 | 哈尔滨工业大学 | A kind of temperature rebuilding method applied to the imaging of flame light field refocusing |
| CN110400336A (en) * | 2019-06-05 | 2019-11-01 | 东南大学 | A kind of double light-field camera flame three dimensional displacement fields methods |
-
2021
- 2021-04-07 CN CN202110376725.0A patent/CN113256812B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608738A (en) * | 2016-03-04 | 2016-05-25 | 华北电力大学（保定） | Light field camera-based flame three-dimensional photometric field reconstruction method |
| CN108389169A (en) * | 2018-03-07 | 2018-08-10 | 哈尔滨工业大学 | A kind of temperature rebuilding method applied to the imaging of flame light field refocusing |
| CN110400336A (en) * | 2019-06-05 | 2019-11-01 | 东南大学 | A kind of double light-field camera flame three dimensional displacement fields methods |
Non-Patent Citations (1)
| Title |
|---|
| 基于单光场相机的火焰三维温度场测量;孙俊等;《工程热物理学报》;20160315(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113256812A (en) | 2021-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111462225B (en) | Centroid identification and positioning method of infrared light spot image | |
| CN1857163B (en) | Methods and apparatus for calibrating CT x-ray beam tracking loop | |
| CN109060836A (en) | High-pressure oil pipe joint external screw thread detection method based on machine vision | |
| CN1903128A (en) | 3D image reconstructing method for a positron CT apparatus, and positron CT apparatus | |
| CN102947727B (en) | By the automatic crystal identification that template realizes can be revised | |
| CN102639990A (en) | High-resolution large-field scanning inspection system for extruded ceramic honeycomb structures | |
| WO2021031711A1 (en) | Method and apparatus for identifying location spectrum, and computer storage medium | |
| CN106547014B (en) | The generation method of crystal localization method and look-up table | |
| CN112924028B (en) | Light field polarization imaging detection system for sea surface oil spill | |
| CN111899344B (en) | Flame emission tomography reconstruction device and method based on camera array | |
| CN113256812B (en) | Characteristic light ray sampling optimization method and processing device for light field camera and storage medium | |
| CN111538149A (en) | Telescope control device, system, method, storage medium, program and application | |
| CN216669794U (en) | Passive nuclear material fast neutron multiplicities measurement system based on spherical symmetrical structure | |
| CN112465952A (en) | Light field camera micro-nano associated imaging sensing-based reconstruction method | |
| CN108874296B (en) | Three-dimensional waveform mapping method based on parallel architecture | |
| CN113237633B (en) | Photoelectric platform stability and precision detection method based on image processing | |
| Chen et al. | An intelligent vision recognition method based on deep learning for pointer meters | |
| CN101907490A (en) | Method for measuring small facula intensity distribution based on two-dimension subdivision method | |
| CN110400336B (en) | Method for reconstructing flame three-dimensional temperature field of double-optical-field camera | |
| CN110646095B (en) | Light field camera radiation sampling detection light processing method | |
| CN109685866A (en) | Crystal positions look-up table construction method, device, computer equipment and storage medium | |
| CN115512343A (en) | Method for correcting and recognizing reading of circular pointer instrument | |
| CN110473261B (en) | Robust light field camera calibration method | |
| CN115755149A (en) | Incident ray intelligent positioning method based on continuous crystal gamma imaging detector | |
| CN211205211U (en) | Hunting bullet outline defect detection device based on laser triangulation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |