CN105141811A - Single-pixel imaging method based on projection coding - Google Patents
Single-pixel imaging method based on projection coding Download PDFInfo
- Publication number
- CN105141811A CN105141811A CN201510226259.2A CN201510226259A CN105141811A CN 105141811 A CN105141811 A CN 105141811A CN 201510226259 A CN201510226259 A CN 201510226259A CN 105141811 A CN105141811 A CN 105141811A
- Authority
- CN
- China
- Prior art keywords
- resolution
- image
- pixel
- matrix
- computer
- 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.)
- Granted
Links
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a single-pixel imaging method based on projection coding and belongs to the digital image processing field. The method comprises steps that, firstly, a portable projector, a computer and a light power meter are utilized to establish a single-pixel imaging platform, and the portable projector is used for flexibly projecting compression sensing coding matrixes generated by the computer, secondly, on the basis, object contour information is rapidly obtained to reduce redundancy calculation in advance, the object contour information is utilized, a crude-to-fine model is employed to guide generation of measurement matrixes, so algorithm complexity of single-pixel imaging recovery is reduced, and the imaging speed of a single-pixel camera is accelerated.
Description
Technical field
The invention belongs to digital image processing field, particularly a kind of portable mono pixel camera imaging platform and the single pixel camera fast imaging method based on hierarchical model and profile information.
Background technology
In the last few years, in signal and image processing field, sparse signal restores problem and receives increasing concern.And compressed sensing this emerging field theoretical is given birth to around this problem just, sparse theory represents when signal is in particular cases sampled at some, just can violate Nyquist law, and obtains the information of abundant signal with less measurement.
The main application direction of compressed sensing theory is exactly incipient single pixel camera imaging technique.In view of its application prospect, single pixel camera application has attracted the interest of more and more people and has been in the center of study hotspot all the time.But prism and aperture are the indispensable parts of this platform, which results in the huge gap between experimental simulation and commercial Application always.
In addition, due to more and more higher to the requirement of image resolution ratio, compressed sensing image taking speed problem is outstanding day by day.Under compressive sensing theory framework, improving image image taking speed has two kinds of modes, and a kind of is optimize final Image Restoration Algorithm, another kind of then be the generation by optimizing calculation matrix, reduction amount of calculation thus accelerate.For the second thinking, most of method now adopts random fashion to produce calculation matrix.The advantage of random fashion is easy realization, but along with the raising of resolution of the image needed, amount of calculation increases substantially, and causes imaging efficiency significantly to decline, and limits the application of single pixel imaging technique to a certain extent.
Summary of the invention
For above-mentioned single pixel camera complex structure, apply dumb and amount of calculation and increase substantially, the technical problem that imaging efficiency significantly declines, the object of this invention is to provide a kind of single pixel formation method based on projection code.
Technical scheme of the present invention is,
Based on a single pixel formation method for projection code, comprise the following steps:
Step one: build single pixel imaging platform;
Single pixel imaging platform comprises portable projector, light power meter and computer, and is fixed together by tripod; Computer for generation of the specified resolution random coded matrix formed by 0,1, and be responsible for data acquisition terminate after image restoration work; The coding pattern that computer produces by portable projector is projected to coding picture realized to image; The light power meter be fixed on projecting apparatus serves as single pixel detector, for obtaining the light intensity of picture reverberation as one-shot measurement result;
Step 2: carry out compressed sensing measurement
For compressive sensing theory: y=Ax, wherein: x column vector representation signal, in practical operation, correspond to the image needing to restore; A represents encoder matrix, in practical operation every a line correspond to computer produce one form random coded matrix by 0,1, the line number of A represents the encoder matrix number of generation; Y column vector represents measurement result, and the reflective light intensity measured each time is as a value of y, and its line number represents pendulous frequency;
Step 3: image restoration, determine finally to need the resolution of the image obtained to be M*N, wherein, M is the line number of required image pixel, and N is the columns of required image pixel; Determine that the image resolution ratio of initial level is m*n, wherein m is the line number of start image pixel, and n is the columns of start image pixel; M, N, m, n are integer; Determine profile threshold value δ;
Step 4: the encoder matrix using Practical computer teaching different resolution, design sketch is shown in Fig. 3, and initial resolution fixes on m*n, thus obtains the random measurement matrix pyramid M of different resolution
1, M
2..., M
k, each different resolution encoder matrix corresponds to the recovery picture I of different resolution
1, I
2..., I
k;
Step 5: utilize l
1-optimization algorithm (can use here
http:// www.acm.caltech.edu/l1magic/the l provided
1-magictoolbox algorithmic tool case) picture of initial resolution m*n is restored, obtain ground floor resolution restored image I
1, loop initialization number of times k=1; l
1-optimization algorithm and L1-norm optimization algorithm.
Step 6: the pixel on the image that traversal has been recovered, obtain the profile information of this image in different resolution, method is: traversal restored image, selects field number of pixels value and is greater than three and pixel value I
ithe profile information G that the pixel that (x, y) is greater than profile threshold value δ is taken turns as i-th
i;
Step 7: calculate the step-length between lower one deck resolution, determine the resolution that lower one deck encoder matrix is corresponding, method is: choose S
i=G
i/ I
ias the adaptive step between i layer and i+1, i.e. step-length S
ithe i-th profile information G taken turns
irestored image pixel value I is taken turns with i-th
iratio;
Step 8: the encoder matrix calculating lower one deck resolution, method is: calculate and upgrade S
i* M
i, i.e. step-length S
iwith i-th layer of resolution matrix M
iproduct as the random coded matrix M of i+1 layer
i+1;
Step 9: utilize M
i+1in conjunction with l
1-optimization algorithm obtains i+1 layer restored image I
i+1, compare the size of its resolution and M*N, if its resolution is more than or equal to M*N, then I
i+1for required restored image, otherwise continue step 6.
Advantageous Effects of the present invention:
The present invention utilizes the simple type of portable projector, can easily realize building of single pixel camera platform, and there will not be problem out of focus owing to not using prism, in addition, utilize contour of object information, and in conjunction with propose by the generation slightly instructing calculation matrix to smart model, thus reduce single pixel imaging recovery algorithm complex, accelerate the image taking speed of single pixel camera.
Accompanying drawing explanation
Fig. 1 is the schematic diagram of single pixel imaging platform.
Fig. 2 is workflow diagram of the present invention.
Fig. 3 is the encoder matrix under the different resolution of computer generation and the encoder matrix after utilizing profile information to optimize
Embodiment
Below, by the invention will be further described with specific embodiment by reference to the accompanying drawings.
Fig. 2 is workflow diagram of the present invention, and a kind of single pixel formation method based on projection code, comprises the following steps:
Single pixel imaging system described in step 1 is built, as shown in Figure 1, and the single pixel imaging system schematic diagram in a specific embodiment of the present invention.The present invention utilizes light power meter, portable projector and computer to form single pixel camera imaging system.Wherein, in compression theory framework, computer for generation of the specified resolution random coded matrix formed by 0,1, and be responsible for data acquisition terminate after image restoration work; The coding pattern that computer produces by portable projector is projected to coding picture realized to image; The light power meter be fixed on projecting apparatus serves as single pixel detector, for obtaining the light intensity of picture reverberation as one-shot measurement result.In the present embodiment, imaging system builds wherein single pixel probe is Newport818-UV/DB, and power counts Newport1916-R-USB, and portable projector looks Q6 for extremely happy, and computer processor is Duo i5-3470, and CPU frequency is 3.20GHz.Distance between equipment can be determined with actual conditions, and this enforcement light power meter is erected at same position above portable projector camera lens, range image 50 centimetres.
According to step 2, carry out compressed sensing measurement.
For compressive sensing theory: y=Ax, x column vector representation signal, in practical operation, correspond to the image needing to restore; A represents encoder matrix, in practical operation every a line correspond to computer produce one form random coded matrix by 0,1, the line number of A represents the encoder matrix number of generation; Y column vector represents measurement result, and the reflective light intensity measured each time is as a value of y, and its line number represents pendulous frequency.
According to step 3, the present embodiment definition is by the image of an acquisition resolution 200*200, and initial pictures resolution is 20*20.The present embodiment has divided 3 levels automatically, determines profile threshold value δ=0.3;
According to step 4 and 5, adopt traditional compressed sensing image recovery method, Fast Restoration goes out the image I of initial resolution 20*20
1, loop initialization number of times k=1; The present invention adopts the tool box (list of references: EmmanuelCandes provided in L1magic, JustinRombergandCaltech.l1-magic:RecoveryofSparseSignals viaConvexProgramming.), the image of the original initial resolution of reflex;
According to step 6, the pixel on the image that traversal has been recovered, obtains the profile information of this image in different resolution, and method is traversal restored image, selects field number of pixels value and is greater than three and pixel value I
ithe profile information G that the pixel that (x, y) is greater than profile threshold value 0.3 is taken turns as i-th
i;
According to step 7, calculate the step-length between lower one deck resolution, determine the resolution that lower one deck encoder matrix is corresponding, method chooses S
i=G
i/ I
ias the adaptive step between i layer and i+1, i.e. step-length S
ithe i-th profile information G taken turns
irestored image pixel value I is taken turns with i-th
iratio;
According to step 8: the encoder matrix calculating lower one deck resolution, method calculates to upgrade S
i* M
i, i.e. step-length S
iwith i-th layer of resolution matrix M
iproduct as the random coded matrix M of i+1 layer
i+1;
According to step 9, utilize M
i+1in conjunction with l
1-optimization algorithm obtains i+1 layer restored image I
i+1, compare the size of its resolution and 200*200, if be more than or equal to, I
i+1for required restored image, otherwise continue step 6.
More than contain the explanation of the preferred embodiment of the present invention; this is to describe technical characteristic of the present invention in detail; be not want summary of the invention to be limited in the concrete form described by embodiment, other amendments carried out according to content purport of the present invention and modification are also protected by this patent.The purport of content of the present invention defined by claims, but not defined by the specific descriptions of embodiment.
Claims (1)
1., based on a single pixel formation method for projection code, it is characterized in that, comprise the following steps:
Step one: build single pixel imaging platform;
Single pixel imaging platform comprises portable projector, light power meter and computer, and is fixed together by tripod; Computer for generation of the specified resolution random coded matrix formed by 0,1, and be responsible for data acquisition terminate after image restoration work; The coding pattern that computer produces by portable projector is projected to coding picture realized to image; The light power meter be fixed on projecting apparatus serves as single pixel detector, for obtaining the light intensity of picture reverberation as one-shot measurement result;
Step 2: carry out compressed sensing measurement
For compressive sensing theory: y=Ax, wherein: x column vector representation signal, in practical operation, correspond to the image needing to restore; A represents encoder matrix, in practical operation every a line correspond to computer produce one form random coded matrix by 0,1, the line number of A represents the encoder matrix number of generation; Y column vector represents measurement result, and the reflective light intensity measured each time is as a value of y, and its line number represents pendulous frequency;
Step 3: image restoration, determine finally to need the resolution of the image obtained to be M*N, wherein, M is the line number of required image pixel, and N is the columns of required image pixel; Determine that the image resolution ratio of initial level is m*n, wherein m is the line number of start image pixel, and n is the columns of start image pixel; M, N, m, n are integer; Determine profile threshold value δ;
Step 4: the encoder matrix using Practical computer teaching different resolution, initial resolution fixes on m*n, thus obtains the random measurement matrix pyramid M of different resolution
1, M
2..., M
k, each different resolution encoder matrix corresponds to the recovery picture I of different resolution
1, I
2..., I
k;
Step 5: utilize the picture of L1-norm optimization algorithm to initial resolution m*n to restore, obtains ground floor resolution restored image I
1, loop initialization number of times k=1;
Step 6: the pixel on the image that traversal has been recovered, obtain the profile information of this image in different resolution, method is: traversal restored image, selects field number of pixels value and is greater than three and pixel value I
ithe profile information G that the pixel that (x, y) is greater than profile threshold value δ is taken turns as i-th
i;
Step 7: calculate the step-length between lower one deck resolution, determine the resolution that lower one deck encoder matrix is corresponding, method is: choose S
i=G
i/ I
ias the adaptive step between i layer and i+1, i.e. step-length S
ithe i-th profile information G taken turns
irestored image pixel value I is taken turns with i-th
iratio;
Step 8: the encoder matrix calculating lower one deck resolution, method is: calculate and upgrade S
i* M
ii.e. step-length S
iwith i-th layer
Resolution matrix M
iproduct as the random coded matrix M of i+1 layer
i+1;
Step 9: utilize M
i+1in conjunction with l
1-optimization algorithm obtains i+1 layer restored image I
i+1, compare the size of its resolution and M*N, if its resolution is more than or equal to M*N, then I
i+1for required restored image, otherwise continue step 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510226259.2A CN105141811B (en) | 2015-05-06 | 2015-05-06 | Based on single pixel formation method of projection code |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510226259.2A CN105141811B (en) | 2015-05-06 | 2015-05-06 | Based on single pixel formation method of projection code |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105141811A true CN105141811A (en) | 2015-12-09 |
| CN105141811B CN105141811B (en) | 2016-04-27 |
Family
ID=54727009
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510226259.2A Active CN105141811B (en) | 2015-05-06 | 2015-05-06 | Based on single pixel formation method of projection code |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105141811B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106887026A (en) * | 2017-01-18 | 2017-06-23 | 四川大学 | Many compression of images and the method rebuild are realized based on compressed sensing and orthogonal modulation |
| CN112153254A (en) * | 2020-08-31 | 2020-12-29 | 合肥工业大学 | Two-step phase-shift single-pixel imaging method based on base map |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104154998A (en) * | 2014-08-15 | 2014-11-19 | 中国科学院上海技术物理研究所 | Reconstruction method for calculating multispectral imaging map based on compressed sensing |
-
2015
- 2015-05-06 CN CN201510226259.2A patent/CN105141811B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104154998A (en) * | 2014-08-15 | 2014-11-19 | 中国科学院上海技术物理研究所 | Reconstruction method for calculating multispectral imaging map based on compressed sensing |
Non-Patent Citations (2)
| Title |
|---|
| 白凌云，梁志毅，徐志军: "基于压缩感知理论的单像素成像系统研究", 《计算机工程与应用》, no. 33, 29 September 2011 (2011-09-29), pages 116 - 119 * |
| 马骏，李少毅，梁志毅，闫杰: "基于压缩传感理论的单像素成像系统设计", 《红外技术》, vol. 33, no. 8, 31 August 2011 (2011-08-31), pages 450 - 456 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106887026A (en) * | 2017-01-18 | 2017-06-23 | 四川大学 | Many compression of images and the method rebuild are realized based on compressed sensing and orthogonal modulation |
| CN106887026B (en) * | 2017-01-18 | 2019-08-16 | 四川大学 | The method for realizing more compression of images and reconstruction based on compressed sensing and orthogonal modulation |
| CN112153254A (en) * | 2020-08-31 | 2020-12-29 | 合肥工业大学 | Two-step phase-shift single-pixel imaging method based on base map |
| CN112153254B (en) * | 2020-08-31 | 2022-02-25 | 合肥工业大学 | Two-step phase-shift single-pixel imaging method based on base map |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105141811B (en) | 2016-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11182917B2 (en) | Stereo camera depth determination using hardware accelerator | |
| JP7161150B2 (en) | Motion compensation for geometry information | |
| Van der Jeught et al. | Real-time structured light profilometry: a review | |
| KR20220029335A (en) | Method and apparatus to complement the depth image | |
| Golla et al. | Real-time point cloud compression | |
| US20200228778A1 (en) | Method and System for Multi-Wavelength Depth Encoding for Three-Dimensional Range Geometry Compression | |
| CN104956404A (en) | Real-time 3d reconstruction with power efficient depth sensor usage | |
| US9467681B2 (en) | Representation and compression of depth data | |
| CN113850900B (en) | Method and system for recovering depth map based on image and geometric clues in three-dimensional reconstruction | |
| CN111612878B (en) | Method and device for making static photo into three-dimensional effect video | |
| CN104157011A (en) | Modeling method for three-dimensional terrain | |
| CN103841406A (en) | Plug and play depth photographic device | |
| TW201436542A (en) | System and method for improving video encoding using content information | |
| CN105141811B (en) | Based on single pixel formation method of projection code | |
| US20230062973A1 (en) | Image processing apparatus, image processing method, and storage medium | |
| CN116468769A (en) | Depth information estimation method based on image | |
| CN113793255A (en) | Method, apparatus, device, storage medium and program product for image processing | |
| Gao et al. | GPU-based real-time structured light 3D scanner at 500 fps | |
| CN108520532B (en) | Method and device for identifying motion direction of object in video | |
| Gu et al. | Ue4-nerf: Neural radiance field for real-time rendering of large-scale scene | |
| US20240054685A1 (en) | Point cloud decoding method, point cloud encoding method, and point cloud decoding device | |
| Garbat et al. | Structured light camera calibration | |
| Daribo et al. | Point cloud compression for grid-pattern-based 3D scanning system | |
| WO2015177183A1 (en) | Method and apparatus for selection of reliable points for 3d modeling | |
| WO2024069727A1 (en) | Training apparatus, training method, and non-transitory computer-readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |