A Joint Modeling Method for Compressed Coding Imaging System

Abstract

The present invention provides a joint modeling method for a compression coding imaging system, comprising: S100, sampling through a compression sampling system, and building a joint model according to the sampling process; S200, obtaining a perception matrix based on the joint model, and applying the perception matrix to perform reconstruction calculation in an implicit expression manner. The use of this solution can reduce the complexity of data storage and calculation.

Description

Technical Field

[0001] The present invention relates to the field of computational imaging, and in particular to a joint modeling method for a compression coding imaging system. Background Art

[0002] In recent years, the requirements for image information acquisition indicators such as spatial, temporal, spectral resolution and sensitivity have gradually increased, which has brought severe challenges to the design and manufacture of optical imaging systems.

[0003] In the field of astronomical observation and remote sensing, in order to achieve higher spatial resolution and sensitivity, optical systems are developing towards a larger aperture, which brings about a sharp increase in technical complexity, construction cost, cycle, and launch cost. At the same time, massive data storage and ground-to-air link transmission are also facing tremendous pressure. In the field of deep space exploration, it is difficult to achieve higher spatial resolution under the premise that resources such as imaging payload mass, volume, and power consumption are strictly limited. In the field of ultrafast imaging and hyperspectral imaging, performance improvement is constrained by factors such as the two-dimensional plane architecture and working frame rate of the detector. Under the traditional imaging system of one-to-one mapping, the performance indicators of the optical system check and balance each other. Improving a certain performance indicator usually requires sacrificing other performance indicators, and technical bottlenecks are beginning to emerge. The compressed sensing theory proposed in recent years has opened up a new technical path to solve the above problems.

[0004] Unlike digital information processing which remains at the mathematical calculation level, compressed imaging systems require the use of physical devices or architectures to complete information collection. When modeling 2D image signals for large-scale images, if the 2D image signals are integrated into the standard compressed sampling model of 1D signals, the corresponding projection matrix and sparse transformation matrix scales will increase exponentially, making data storage and calculation more difficult. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the prior art and to propose a joint modeling method for a compression coding imaging system, which can reduce the complexity of data storage and calculation.

[0006] To achieve the above object, the present invention adopts the following specific technical solutions:

[0007] A joint modeling method for a compression coding imaging system according to an embodiment of the present invention includes:

[0008] S100, sampling is performed through a compression sampling system, and a joint model is constructed according to the sampling process;

[0009] S200, obtaining a perception matrix based on the joint model, and applying the perception matrix to perform reconstruction calculation in an implicit expression manner.

[0010] The present invention can at least achieve the following beneficial effects: based on the hardware architecture of the compressed imaging system, the present solution unifies the sampling process and results into a joint model, thereby making full use of the sampling process and the sparse properties of the 2D image, effectively improving the image reconstruction quality, and reducing the complexity of data storage and calculation. At the same time, through the implicit expression and implicit operation of the projection matrix in the joint model, the image reconstruction process is accelerated, further reducing the complexity of data calculation.

[0011] According to some embodiments of the present invention, step S100 includes:

[0012] S110, performing compression sampling in orthogonal directions through a compression sampling system;

[0013] S120, constructing a model according to the sampling process;

[0014] S130, performing sparse transformation on the models and jointly expressing the models as the joint model.

[0015] According to some embodiments of the present invention, the model process is expressed as:

[0016] ;

[0017] where Y V , Y H They are the detector sampling processes in the vertical and horizontal directions, A1 and A2 are the projection matrices corresponding to the TDI charge transfer and coded exposure processes, X is the target image matrix, and B is the pixel merging matrix.

[0018] According to some embodiments of the present invention, the TDI detector is a three-phase TDI detector.

[0019] According to some embodiments of the present invention, the projection matrix A1 is:

[0020]

[0021] Where S={S i , i∈[1,3×(M-1)+N+2]} is the coded exposure sequence, N is the detector TDI integration level, M is the number of sampling rows for each detector,

[0022] The projection matrix A2 is the projection matrix A1 rearranged in reverse order by columns, that is, A2=fliplr(A1).

[0023] According to some embodiments of the present invention, the structure of the pixel merging matrix B is expressed as:

[0024] .

[0025] According to some embodiments of the present invention, step S130 includes: H Transpose both sides of the equation and perform sparse transformation on the target image matrix X. The model is shown in the following formula after sorting:

[0026] ;

[0027] Where ψ and φ are sparse transformation matrices in two orthogonal directions, respectively, and the matrix θ is the expression of the target image matrix X in the sparse domain.

[0028] According to some embodiments of the present invention, step S200 includes:

[0029] ;

[0030] in represents the Kronecker product operation, Vec(·) represents the row-wise vectorization of the matrix, and Λ is the perception matrix.

[0031] According to some embodiments of the present invention, step S200 further includes: calculating a product vector R1 of the perception matrix Λ and an arbitrary vector Vec(z) by the following formula:

[0032] ;

[0033] Mat(·) represents the matrix processing of vectors arranged in rows, R 1up Represents the first half of the result vector R1, R 1down Represents the second half of the result vector R1.

[0034] According to some embodiments of the present invention, step S200 further includes:

[0035] For the transpose of the perception matrix Λ T The product vector R2 of any vector Vec(z) is calculated by the following formula:

[0036] ;

[0037] Where Vec(z) up Represents the first half of any vector Vec(z), Vec(z) down Represents the second half of any vector Vec(z).

[0038] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0040] Figure 1 is a schematic diagram of a compression coding imaging system according to an embodiment of the present invention;

[0041] Figure 2 It is a comparison of reconstructed image curves of the joint modeling method of the compression coding imaging system according to the embodiment of the invention and the existing method;

[0042] Figure 3 is a flow chart of a joint modeling method for a compression coding imaging system according to an embodiment of the present invention;

[0043] Figure 4 is a flow chart of S100 according to an embodiment of the present invention.

[0044] Reference numerals include:

[0045] Target scene 1, lens 2, beam splitter 3, reflector 4, detector 5, control and data acquisition circuit 6, host computer 7. DETAILED DESCRIPTION

[0046] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the following description, the same modules are represented by the same reference numerals. In the case of the same reference numerals, their names and functions are also the same. Therefore, the detailed description thereof will not be repeated.

[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation of the present invention.

[0048] To achieve the above object, the present invention adopts the following specific technical solutions:

[0049] According to the joint modeling method of the compression coding imaging system of the embodiment of the present invention, Figure 3 As shown, including:

[0050] S100, sampling is performed through a compression sampling system, and a joint model is constructed according to the sampling process;

[0051] S200, obtaining a perception matrix based on the joint model, and applying the perception matrix to perform reconstruction calculation in an implicit expression manner.

[0052] Compression coding imaging system such as Figure 1As shown, the light source of the target scene 1 passes through the lens 2 and is divided into two paths by the beam splitter 3, one of which passes through the lens 2 and is received by the control and data acquisition circuit 6, and the other passes through the lens 2 after the light path is changed by the reflector 2 and is received by the control and data acquisition circuit 6. The signals collected by the two control and data acquisition circuits 6 are orthogonal to each other, and the control and data acquisition circuit 6 transmits the received data to the host computer 7 for processing.

[0053] Based on the hardware architecture of the compressed imaging system, the sampling process and results are unified into a joint model, which fully utilizes the sampling process and the sparse properties of the 2D image, effectively improves the image reconstruction quality, and reduces the complexity of data storage and calculation. At the same time, through the implicit expression and implicit operation of the projection matrix in the joint model, the image reconstruction process is accelerated, further reducing the complexity of data calculation.

[0054] According to some embodiments of the present invention, Figure 4 As shown, step S100 includes:

[0055] S110, performing compression sampling in orthogonal directions through a compression sampling system;

[0056] S120, constructing a model according to the sampling process;

[0057] S130, performing sparse transformation on the models and jointly expressing the models as a joint model.

[0058] In existing processing methods, the sparse properties of 2D images are not fully utilized, and the sampling processes in two directions are not jointly modeled, resulting in low efficiency in extracting sampling information.

[0059] The sampling processes and results in two orthogonal directions are unified into a joint model, thereby making full use of the sampling processes in two orthogonal directions and the sparse properties of 2D images, effectively improving the image reconstruction quality and reducing the complexity of data storage and calculation.

[0060] According to some embodiments of the present invention, the process of step S110 is expressed as:

[0061] ;

[0062] where Y V , Y H They are the detector sampling processes in the vertical and horizontal directions, A1 and A2 are the projection matrices corresponding to the TDI charge transfer and coded exposure processes, X is the target image matrix, and B is the pixel merging matrix.

[0063] Sampling process Y V , Y H is the detector sampling process in the vertical and horizontal directions, YV , Y H Mutually orthogonal. Use A1, A2, X, B to sample process Y V , Y H To express.

[0064] According to some embodiments of the present invention, the TDI detector is a three-phase TDI detector.

[0065] When different TDI detectors are used for detection, the projection matrices A1, A2 and pixel merging matrix B corresponding to the charge transfer and coding exposure processes are different.

[0066] According to some embodiments of the present invention, when a three-phase TDI detector is used, the projection matrix A1 is:

[0067]

[0068] Where S={S i , i∈[1,3×(M-1)+N+2]} is the coded exposure sequence, N is the detector TDI integration level, M is the number of sampling rows for each detector, and since Y V , Y H The directions are orthogonal to each other, so the projection matrix A2 is the projection matrix A1 rearranged in reverse order by columns, that is, A2=fliplr(A1).

[0069] According to some embodiments of the present invention, when a three-phase TDI detector is used, the structure of the pixel merging matrix B is expressed as:

[0070] .

[0071] Among them, the size of matrix B is 3N rows and N columns.

[0072] According to some embodiments of the present invention, step S130 includes: H Transpose both sides of the equation and perform sparse transformation on the target image matrix X. The model is shown in the following formula after sorting:

[0073] ;

[0074] Where ψ and φ are sparse transformation matrices in two orthogonal directions, respectively, and the matrix θ is the expression of the target image matrix X in the sparse domain.

[0075] Perform a sparse transformation on the target image matrix X to fully utilize the sparse properties of the 2D image. ψ and φ can be the same or different.

[0076] According to some embodiments of the present invention, step S200 includes:

[0077] ;

[0078] in represents the Kronecker product operation, Vec(·) represents the row vectorization of the content matrix in the brackets, Λ is the perception matrix, ψ T ,φ T , Y H T , B T is ψ, φ, Y H , the transposed matrix of B.

[0079] The joint model is processed to obtain the perception matrix. Calculating the implicit expression of the perception matrix can reduce the amount of calculation in the process of reconstructing the compressed image. For most compressed sampling reconstruction algorithms, the reconstruction process involves two types of operations on the perception matrix Λ: the product of Λ and an arbitrary vector, and the product of Λ transpose and an arbitrary vector.

[0080] According to some embodiments of the present invention, step S200 further includes: calculating the product result vector R1 of the perception matrix Λ and the arbitrary vector Vec(z) by the following formula:

[0081] ;

[0082] Mat(·) represents the matrix processing of vectors arranged in rows, R 1up Represents the first half of the result vector R1, R 1down Represents the second half of the result vector R1.

[0083] According to some embodiments of the present invention, step S200 further includes:

[0084] For the transpose of the perception matrix Λ T The product vector R2 of any vector Vec(z) is calculated by the following formula:

[0085] ;

[0086] Where Vec(z) up Represents the first half of any vector Vec(z), Vec(z) down Represents the second half of any vector Vec(z).

[0087] The product vector R1 of the perception matrix Λ and any vector Vec(z) and the transpose Λ of the perception matrix T The product vector R2 calculated with any vector Vec(z) can satisfy the reconstruction process of most algorithms.

[0088] The reconstructed image curves of the joint modeling method of the compression coding imaging system of this scheme are compared with those of the existing methods. Figure 2As shown, it can be found that the image quality of this method is higher.

[0089] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0090] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

[0091] The above specific implementations of the present invention do not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A joint modeling method for a compression coding imaging system, characterized in that: include: S100, sampling is performed through a compression sampling system, and a joint model is constructed according to the sampling process; The step S100 includes: S110, performing compression sampling in orthogonal directions through a compression sampling system; S120, constructing a model according to the sampling process; The model is expressed as: ； where Y V , Y H are the detector sampling processes in the vertical and horizontal directions, respectively. A1 and A2 are the projection matrices corresponding to the TDI charge transfer and coded exposure processes. X is the target image matrix, and B is the pixel merging matrix. S130, performing sparse transformation on the models and jointly expressing the models as the joint model; S200, obtaining a perception matrix based on the joint model, and applying the perception matrix to perform reconstruction calculation in an implicit expression manner.

2. The joint modeling method of the compression coding imaging system according to claim 1, characterized in that: The TDI detector is a three-phase TDI detector.

3. The joint modeling method of the compression coding imaging system according to claim 2, characterized in that: The projection matrix A1 is: ； Where S={S i , i∈[1,3×(M-1)+N+2]} is the coded exposure sequence, N is the detector TDI integration level, M is the number of sampling rows for each detector, The projection matrix A2 is the projection matrix A1 rearranged in reverse order by columns, that is, A2=fliplr(A1).

4. The joint modeling method of the compression coding imaging system according to claim 2, characterized in that: The structural form of the pixel merging matrix B is expressed as: 。 5. The joint modeling method of the compression coding imaging system according to claim 1, characterized in that: The step S130 includes: Y H Transpose both sides of the equation and perform sparse transformation on the target image matrix X. The model is shown in the following formula after sorting: ； Where ψ and φ are sparse transformation matrices in two orthogonal directions, respectively, and the matrix θ is the expression of the target image matrix X in the sparse domain.

6. The joint modeling method of the compression coding imaging system according to claim 5, characterized in that: The step S200 includes: ； in represents the Kronecker product operation, Vec(·) represents the row-wise vectorization of the matrix, and Λ is the perception matrix.

7. The joint modeling method of the compression coding imaging system according to claim 6, characterized in that: The step S200 further includes: The product vector R1 of the perception matrix Λ and any vector Vec(z) is calculated by the following formula: ； Mat(·) represents the matrix processing of vectors arranged in rows, R 1up Represents the first half of the result vector R1, R 1down Represents the second half of the result vector R1.

8. The joint modeling method of the compression coding imaging system according to claim 7, characterized in that: The step S200 further includes: For the transpose of the perception matrix Λ T The product vector R2 of any vector Vec(z) is calculated by the following formula: ； Where Vec(z) up Represents the first half of any vector Vec(z), Vec(z) down Represents the second half of any vector Vec(z).

Images