CN107680146A - Method for reconstructing, device, equipment and the storage medium of PET image - Google Patents

Method for reconstructing, device, equipment and the storage medium of PET image Download PDF

Info

Publication number
CN107680146A
CN107680146A CN201710821065.6A CN201710821065A CN107680146A CN 107680146 A CN107680146 A CN 107680146A CN 201710821065 A CN201710821065 A CN 201710821065A CN 107680146 A CN107680146 A CN 107680146A
Authority
CN
China
Prior art keywords
pet image
resolution
low
image
resolution pet
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.)
Pending
Application number
CN201710821065.6A
Other languages
Chinese (zh)
Inventor
胡战利
李涛
杨永峰
梁栋
刘新
郑海荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to CN201710821065.6A priority Critical patent/CN107680146A/en
Publication of CN107680146A publication Critical patent/CN107680146A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/003—Reconstruction from projections, e.g. tomography
    • G06T11/008—Specific post-processing after tomographic reconstruction, e.g. voxelisation, metal artifact correction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2211/00—Image generation
    • G06T2211/40—Computed tomography
    • G06T2211/416—Exact reconstruction

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)

Abstract

本发明适用计算机技术领域，提供了一种PET图像的重建方法、装置、设备以及存储介质，该方法包括：对接收的低分辨率PET图像进行分割，根据分割后得到的每个低分辨率PET图像块、低分辨率字典和高分辨率字典，生成每个低分辨率PET图像块的稀疏系数，根据这些稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像，根据低分辨率PET图像、高分辨率PET图像、模糊矩阵和下采样矩阵，生成并输出低分辨率PET图像的重建图像，从而有效地降低了PET图像重建的时间消耗，有效地提高了PET图像重建效率和PET图像重建后的图像质量。

The present invention is applicable to the field of computer technology, and provides a PET image reconstruction method, device, equipment, and storage medium. The method includes: segmenting the received low-resolution PET image, and according to each low-resolution PET image obtained after segmentation, Image block, low-resolution dictionary and high-resolution dictionary, generate sparse coefficients for each low-resolution PET image block, and generate high-resolution PET images corresponding to low-resolution PET images based on these sparse coefficients and high-resolution dictionaries, According to the low-resolution PET image, high-resolution PET image, blur matrix and downsampling matrix, the reconstructed image of the low-resolution PET image is generated and output, thereby effectively reducing the time consumption of PET image reconstruction and effectively improving the PET image Reconstruction efficiency and image quality after PET image reconstruction.

Description

PET图像的重建方法、装置、设备及存储介质PET image reconstruction method, device, equipment and storage medium

技术领域technical field

本发明属于图像处理技术领域，尤其涉及一种PET图像的重建方法、装置、设备及存储介质。The invention belongs to the technical field of image processing, and in particular relates to a PET image reconstruction method, device, equipment and storage medium.

背景技术Background technique

正电子发射断层扫描成像(PET)具有很高的临床价值，但是由于PET成像原理以及PET硬件系统的限制，正常扫描得到的PET图像往往会很模糊，而且会丢失部分边缘信息。用扫描得到的低剂量采样数据进行PET图像重建，重建得到的PET图像的图像质量会更差，包含大量的噪声，甚至产生图像伪影。Positron emission tomography (PET) has high clinical value, but due to the limitations of PET imaging principles and PET hardware systems, PET images obtained by normal scanning are often blurred and part of the edge information will be lost. Using the low-dose sampling data obtained from scanning for PET image reconstruction, the image quality of the reconstructed PET image will be worse, contain a lot of noise, and even produce image artifacts.

目前，传统的PET重建算法在抑制图像噪声、修复图像边缘信息方面存在较多的限制，无法有效地提高PET图像质量，现有对PET图像进行重建的算法大多采用迭代优化的方式来提高PET图像的质量，不仅会耗费较长的时间，而且通过迭代优化的方式对欠采样数据进行图像伪影的抑制，会使得图像丢失一些细节特征。At present, traditional PET reconstruction algorithms have many limitations in suppressing image noise and repairing image edge information, and cannot effectively improve the quality of PET images. Most of the existing reconstruction algorithms for PET images use iterative optimization to improve PET image It will not only take a long time, but also suppress the image artifacts of the undersampled data through iterative optimization, which will cause the image to lose some detailed features.

发明内容Contents of the invention

本发明的目的在于提供一种PET图像的重建方法、装置、设备及存储介质，旨在解决现有技术中PET图像重建的效率不高，且重建得到的图像质量不佳的问题。The object of the present invention is to provide a PET image reconstruction method, device, equipment and storage medium, aiming to solve the problems of low PET image reconstruction efficiency and poor quality of reconstructed images in the prior art.

一方面，本发明提供了一种PET图像的重建方法，所述方法包括下述步骤：On the one hand, the present invention provides a kind of reconstruction method of PET image, and described method comprises the following steps:

接收用户输入的低分辨率PET图像，对所述低分辨率PET图像进行分割，生成低分辨率PET图像块；receiving a low-resolution PET image input by a user, and segmenting the low-resolution PET image to generate low-resolution PET image blocks;

根据所述每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成所述每个低分辨率PET图像块的稀疏系数；According to each low-resolution PET image block, pre-trained low-resolution dictionary and high-resolution dictionary, generate the sparse coefficient of each low-resolution PET image block;

根据所述每个低分辨率PET图像块的稀疏系数和所述高分辨率字典，生成所述低分辨率PET图像对应的高分辨率PET图像；Generate a high-resolution PET image corresponding to the low-resolution PET image according to the sparse coefficient of each low-resolution PET image block and the high-resolution dictionary;

根据所述低分辨率PET图像、所述高分辨率PET图像、预设的模糊矩阵和预设的下采样矩阵，生成并输出所述低分辨率PET图像的重建图像。Generate and output a reconstructed image of the low-resolution PET image according to the low-resolution PET image, the high-resolution PET image, a preset blur matrix and a preset downsampling matrix.

另一方面，本发明提供了一种PET图像的重建装置，所述装置包括：In another aspect, the present invention provides a PET image reconstruction device, the device comprising:

图像分割单元，用于接收用户输入的低分辨率PET图像，对所述低分辨率PET图像进行分割，生成低分辨率PET图像块；An image segmentation unit, configured to receive a low-resolution PET image input by a user, segment the low-resolution PET image, and generate low-resolution PET image blocks;

系数生成单元，用于根据所述每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成所述每个低分辨率PET图像块的稀疏系数；A coefficient generation unit, configured to generate sparse coefficients for each low-resolution PET image block according to each low-resolution PET image block, a pre-trained low-resolution dictionary and a high-resolution dictionary;

图像生成单元，用于根据所述每个低分辨率PET图像块的稀疏系数和所述高分辨率字典，生成所述低分辨率PET图像对应的高分辨率PET图像；以及An image generating unit, configured to generate a high-resolution PET image corresponding to the low-resolution PET image according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary; and

图像输出单元，用于根据所述低分辨率PET图像、所述高分辨率PET图像、预设的模糊矩阵和预设的下采样矩阵，生成并输出所述低分辨率PET图像的重建图像。An image output unit, configured to generate and output a reconstructed image of the low-resolution PET image according to the low-resolution PET image, the high-resolution PET image, a preset blur matrix and a preset downsampling matrix.

另一方面，本发明还提供了一种医学图像处理设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现如上述PET图像的重建方法所述的步骤。On the other hand, the present invention also provides a medical image processing device, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, and the processor executes the computer program At the same time, the steps described in the reconstruction method of the above-mentioned PET image are realized.

另一方面，本发明还提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行时实现如上述PET图像的重建方法所述的步骤。On the other hand, the present invention also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps described in the above PET image reconstruction method are realized. .

本发明根据分割后的低分辨率PET图像块、训练好的低分辨率字典和高分辨率字典，计算每个低分辨率PET图像块的稀疏系数，根据这些稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，通过对高分辨率PET图像进行后处理，得到低分辨率PET的重建图像，从而有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建中的计算复杂度，进而有效地提高了PET图像的重建效率。The present invention calculates the sparse coefficient of each low-resolution PET image block according to the segmented low-resolution PET image block, the trained low-resolution dictionary and the high-resolution dictionary, and calculates the low-resolution PET image block according to these sparse coefficients and the high-resolution dictionary The high-resolution PET image corresponding to the high-resolution PET image, by post-processing the high-resolution PET image, obtains the reconstructed image of the low-resolution PET, thereby effectively improving the image quality of the PET image reconstruction and effectively reducing the PET image Computational complexity in reconstruction, thus effectively improving the reconstruction efficiency of PET images.

附图说明Description of drawings

图1是本发明实施例一提供的PET图像的重建方法的实现流程图；Fig. 1 is the implementation flowchart of the reconstruction method of PET image provided by Embodiment 1 of the present invention;

图2是本发明实施例二提供的PET图像的重建方法的实现流程图；Fig. 2 is the implementation flowchart of the PET image reconstruction method provided by Embodiment 2 of the present invention;

图3是本发明实施例三提供的PET图像的重建装置的结构示意图；FIG. 3 is a schematic structural diagram of a PET image reconstruction device provided by Embodiment 3 of the present invention;

图4是本发明实施例四提供的PET图像的重建装置的结构示意图；以及FIG. 4 is a schematic structural diagram of a PET image reconstruction device provided by Embodiment 4 of the present invention; and

图5是本发明实施例五提供的医学图像处理设备的结构示意图。Fig. 5 is a schematic structural diagram of a medical image processing device provided by Embodiment 5 of the present invention.

具体实施方式detailed description

为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

以下结合具体实施例对本发明的具体实现进行详细描述：The specific realization of the present invention is described in detail below in conjunction with specific embodiment:

实施例一：Embodiment one:

图1示出了本发明实施例一提供的PET图像的重建方法的实现流程，为了便于说明，仅示出了与本发明实施例相关的部分，详述如下：Figure 1 shows the implementation process of the PET image reconstruction method provided by Embodiment 1 of the present invention. For the convenience of description, only the parts related to the embodiment of the present invention are shown, and the details are as follows:

在步骤S101中，接收用户输入的低分辨率PET图像，对低分辨率PET图像进行分割，生成低分辨率PET图像块；In step S101, the low-resolution PET image input by the user is received, the low-resolution PET image is segmented, and the low-resolution PET image block is generated;

本发明实施例适用于正电子发射断层成像(PET)图像的图像重建。可通过预设的图像分割算法对接收到的用户输入的低分辨率PET图像进行图像块的分割，生成多个低分辨率PET图像块，具体地，在分割时每个低分辨率PET图像块的尺寸相同，且前后相邻的低分辨率PET图像块之间存在重叠区域。Embodiments of the present invention are applicable to image reconstruction of Positron Emission Tomography (PET) images. The received user-input low-resolution PET image can be divided into image blocks by a preset image segmentation algorithm to generate multiple low-resolution PET image blocks. Specifically, each low-resolution PET image block have the same size, and there is an overlapping area between adjacent low-resolution PET image blocks.

在步骤S102中，根据每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成每个低分辨率PET图像块的稀疏系数。In step S102, generate sparse coefficients for each low-resolution PET image block according to each low-resolution PET image block, the pre-trained low-resolution dictionary and the high-resolution dictionary.

在本发明实施例中，低分辨率字典和高分辨率字典为预先训练好的过完备字典，低分辨率PET图像块的稀疏系数即低分辨率PET图像块的稀疏表示，低分辨率PET图像块对应的高分辨率PET图像块可近似为高分辨率字典与低分辨率PET图像块的稀疏系数的乘积，因此需对低分辨率PET图像块的稀疏系数进行求解，以进一步得到高分辨率PET图像块。In the embodiment of the present invention, the low-resolution dictionary and the high-resolution dictionary are pre-trained overcomplete dictionaries, and the sparse coefficients of the low-resolution PET image block are the sparse representation of the low-resolution PET image block, and the low-resolution PET image block The high-resolution PET image block corresponding to the block can be approximated as the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block, so it is necessary to solve the sparse coefficient of the low-resolution PET image block to further obtain the high-resolution PET image blocks.

在本发明实施例中，低分辨率PET图像块的稀疏系数可通过下列公式求解：In the embodiment of the present invention, the sparse coefficient of the low-resolution PET image block can be solved by the following formula:

min||α||0，且其中，F为预设的特征提取函数，α为低分辨率PET图像块y的稀疏系数，D1为低分辨率字典，ε为预设阈值。该公式求解出的稀疏系数结合低分辨率字典可精确地表示出低分辨率PET图像块，但该公式的求解是个NP-hard问题，可将该公式等效为求解L1范式最小化的过程：min||α|| 0 , and Among them, F is the preset feature extraction function, α is the sparse coefficient of the low-resolution PET image block y, D 1 is the low-resolution dictionary, and ε is the preset threshold. The sparse coefficients solved by this formula combined with the low-resolution dictionary can accurately represent low-resolution PET image blocks, but the solution of this formula is an NP-hard problem, and the formula can be equivalent to the process of solving the minimization of L1 normal form:

其中，λ为预设参数，用来平衡α的稀疏性和α与y的保真度。为了使得稀疏系数α的求解更为精确，即使得通过稀疏系数α恢复得到的高分辨率PET图像块与对应的低分辨率PET图像块的相关度更高，在这里对公式进行了更为准确的计算： Among them, λ is a preset parameter, which is used to balance the sparsity of α and the fidelity of α and y. In order to make the solution of the sparse coefficient α more accurate, that is, to make the high-resolution PET image block restored by the sparse coefficient α have a higher correlation with the corresponding low-resolution PET image block, the formula A more accurate calculation is made:

min||α||1，其中，α要求满足第一系数约束条件和第二系数约束条件其中，ε1为预设的第一阈值，ε2为预设的第二阈值，P用来提取当前低分辨率PET图像块与上一低分辨率PET图像块的重叠区域，ω为上一低分辨率PET图像块在重叠区域的值。最终，计算得到每块低分辨率PET图像块的稀疏系数。min||α|| 1 , where α is required to satisfy the first coefficient constraint and the second coefficient constraint Among them, ε1 is the preset first threshold, ε2 is the preset second threshold, P is used to extract the overlapping area between the current low-resolution PET image block and the previous low-resolution PET image block, and ω is the previous low-resolution PET image block. Values of low-resolution PET image patches in overlapping regions. Finally, the sparse coefficient of each low-resolution PET image block is calculated.

在步骤S103中，根据每个低分辨率PET图像块的稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像。In step S103, according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary, a high-resolution PET image corresponding to the low-resolution PET image is generated.

在本发明实施例中，根据每个低分辨率PET图像块的稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像，具体地，每块低分辨率PET图像块对应的高分辨率PET图像可通过公式x＝D2α计算得到，其中，x为高分辨率PET图像块。In the embodiment of the present invention, according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary, a high-resolution PET image corresponding to the low-resolution PET image is generated, specifically, each low-resolution PET image block The corresponding high-resolution PET image can be calculated by the formula x=D 2 α, where x is a high-resolution PET image block.

在步骤S104中，根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵和预设的下采样矩阵，生成并输出低分辨率PET图像的重建图像。In step S104, a reconstructed image of the low-resolution PET image is generated and output according to the low-resolution PET image, the high-resolution PET image, the preset blur matrix and the preset down-sampling matrix.

在本发明实施例中，由于在求解低分辨率PET图像块的稀疏系数时采用了近似逼近的方式，可能使得稀疏系数的准确度较低，求得的高分辨率PET图像质量不佳，因此生成并输出低分辨率PET图像的重建图像时需要对高分辨率PET图像进行优化。为了便于描述，将高分辨率PET图像设置为低分辨率PET图像的第一重建图像，根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成第二重建图像，其中，梯度下降方式可表示为：In the embodiment of the present invention, due to the use of an approximate approximation method when solving the sparse coefficients of low-resolution PET image blocks, the accuracy of the sparse coefficients may be low, and the obtained high-resolution PET image quality is not good, so Generating and exporting reconstructions of low-resolution PET images requires optimization of high-resolution PET images. For the convenience of description, the high-resolution PET image is set as the first reconstructed image of the low-resolution PET image, according to the low-resolution PET image, high-resolution PET image, preset blur matrix, preset downsampling matrix and preset The gradient descent method is set to optimize the first reconstructed image to generate the second reconstructed image, where the gradient descent method can be expressed as:

Xt+1＝Xt+v[HTST(Y-SHXt)+c(Xt-X0)]，其中，Xt为第t次优化过程中的第一重建图像，Xt+1为第t次优化过程中的第二重建图像，v为预设的梯度下降步长，H为模糊矩阵，S为矢量矩阵，Y为低分辨率PET图像，X0为高分辨率PET图像，c为预设参数。判断当前优化次数t是否达到预设的最大优化次数，当达到时，输出第二重建图像，否则，将第二重建图像设置为第一重建图像，对当前优化次数t进行加一操作，并跳转至执行根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化的步骤，从而通过对高分辨率PET图像进行优化，有效地提高了低分辨率PET图像重建的质量，并有效地降低了低分辨率PET图像重建的计算复杂度。X t+1 =X t +v[H T S T (Y-SHX t )+c(X t -X 0 )], where X t is the first reconstructed image in the t-th optimization process, and X t +1 is the second reconstructed image in the t-th optimization process, v is the preset gradient descent step size, H is the blur matrix, S is the vector matrix, Y is the low-resolution PET image, X 0 is the high-resolution PET image, c is a preset parameter. Judging whether the current optimization times t has reached the preset maximum number of optimization times, when reached, output the second reconstructed image, otherwise, set the second reconstructed image as the first reconstructed image, add one to the current optimization times t, and skip Go to the step of optimizing the first reconstructed image according to the low-resolution PET image, the high-resolution PET image, the preset blur matrix, the preset down-sampling matrix and the preset gradient descent method, thereby by The optimization of high-resolution PET images effectively improves the quality of low-resolution PET image reconstruction and effectively reduces the computational complexity of low-resolution PET image reconstruction.

在本发明实施例中，结合训练好的低分辨率字典和高分辨率字典构造第一系数约束条件和第二系数约束条件，根据第一系数约束条件和第二系数约束条件，对分割后的低分辨率PET图像块对应的稀疏系数进行近似计算，有效地提高低分辨率PET图像稀疏系数的准确度，根据稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，并通过梯度下降方式对高分辨率PET图像进行优化，有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建的计算复杂度，从而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the first coefficient constraint condition and the second coefficient constraint condition are constructed in combination with the trained low-resolution dictionary and high-resolution dictionary, and according to the first coefficient constraint condition and the second coefficient constraint condition, the divided The sparse coefficient corresponding to the low-resolution PET image block is approximated to effectively improve the accuracy of the low-resolution PET image sparse coefficient, and the high-resolution PET image corresponding to the low-resolution PET image is calculated according to the sparse coefficient and the high-resolution dictionary. And the high-resolution PET image is optimized by gradient descent, which effectively improves the image quality of PET image reconstruction, effectively reduces the computational complexity of PET image reconstruction, and thus effectively improves the reconstruction efficiency of PET image.

实施例二：Embodiment two:

图2示出了本发明实施例二提供的PET图像的重建方法的实现流程，为了便于说明，仅示出了与本发明实施例相关的部分，详述如下：Fig. 2 shows the implementation process of the PET image reconstruction method provided by the second embodiment of the present invention. For the convenience of description, only the parts related to the embodiment of the present invention are shown, and the details are as follows:

在步骤S201中，对低分辨率字典和高分辨率字典进行随机初始化。In step S201, the low-resolution dictionary and the high-resolution dictionary are randomly initialized.

在步骤S202中，根据预设的低分辨率PET训练图集、预设的高分辨率PET训练图集、低分辨率PET训练图集中图像块的尺寸、以及高分辨率PET训练图集中图像块的尺寸，对低分辨率字典和高分辨率字典进行联合训练。In step S202, according to the preset low-resolution PET training atlas, the preset high-resolution PET training atlas, the size of the image block in the low-resolution PET training atlas, and the image block in the high-resolution PET training atlas The size of the low-resolution dictionary and the high-resolution dictionary are jointly trained.

在本发明实施例中，可通过高斯随机矩阵对低分辨率字典和高分辨率字典进行随机初始化，由于高分辨字典与低分辨率PET图像块的稀疏系数的乘积可近似为高分辨率PET图像块，而低分辨率PET图像块的稀疏系数与低分辨率字典的乘积为低分辨率PET图像块，可看出高分辨率字典与低分辨率字典间具有联系，因此需对低分辨率字典和高分辨率字典进行联合训练。具体地，联合训练的公式为：In the embodiment of the present invention, the low-resolution dictionary and the high-resolution dictionary can be randomly initialized by a Gaussian random matrix, because the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block can be approximated as a high-resolution PET image block, and the product of the sparse coefficient of the low-resolution PET image block and the low-resolution dictionary is the low-resolution PET image block. It can be seen that there is a connection between the high-resolution dictionary and the low-resolution dictionary, so the low-resolution dictionary Joint training with high-resolution dictionaries. Specifically, the formula for joint training is:

其中，c＝1,2，X1为预设的低分辨率PET训练图集中低分辨率PET训练图像，X2为预设的高分辨率PET训练图集中高分辨率PET训练图像，Z为预设的矩阵变量，N为低分辨率PET训练图像的图像块尺寸，M为高分辨率PET训练图像的图像块尺寸。 Among them, c=1,2, X 1 is the low-resolution PET training image in the preset low-resolution PET training atlas, X 2 is the high-resolution PET training image in the preset high-resolution PET training atlas, and Z is Preset matrix variables, N is the image block size of the low-resolution PET training image, and M is the image block size of the high-resolution PET training image.

在步骤S203中，接收用户输入的低分辨率PET图像，对低分辨率PET图像进行分割，生成低分辨率PET图像块。In step S203, the low-resolution PET image input by the user is received, and the low-resolution PET image is segmented to generate low-resolution PET image blocks.

在本发明实施例中，可通过预设的图像分割算法对接收到的用户输入的低分辨率PET图像进行图像块的分割，生成低分辨率PET图像块，每个低分辨率PET图像块的尺寸相同，前后相邻的低分辨率PET图像块之间存在重叠区域。In the embodiment of the present invention, the low-resolution PET image received by the user may be segmented into image blocks by a preset image segmentation algorithm to generate low-resolution PET image blocks, and each low-resolution PET image block The size is the same, and there is an overlapping area between the adjacent low-resolution PET image blocks.

在步骤S204中，根据每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成每个低分辨率PET图像块的稀疏系数。In step S204, generate sparse coefficients for each low-resolution PET image block according to each low-resolution PET image block, the pre-trained low-resolution dictionary and the high-resolution dictionary.

在本发明实施例中，低分辨率PET图像块对应的高分辨率PET图像块可近似为高分辨率字典与低分辨率PET图像块的稀疏系数的乘积，因此需对低分辨率PET图像块的稀疏系数进行求解，以进一步得到高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image block corresponding to the low-resolution PET image block can be approximated as the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block, so the low-resolution PET image block needs to be The sparse coefficients are solved to further obtain high-resolution PET image blocks.

在本发明实施例中，低分辨率PET图像块的稀疏系数可通过下列公式求解：In the embodiment of the present invention, the sparse coefficient of the low-resolution PET image block can be solved by the following formula:

min||α||0，且其中，F为预设的特征提取函数，α为低分辨率PET图像块y的稀疏系数，D1为低分辨率字典，ε为预设阈值。该公式求解出的稀疏系数结合低分辨率字典可精确地表示出低分辨率PET图像块，但该公式的求解是个NP-hard问题，可将该公式等效为求解L1范式最小化的过程：min||α|| 0 , and Among them, F is the preset feature extraction function, α is the sparse coefficient of the low-resolution PET image block y, D 1 is the low-resolution dictionary, and ε is the preset threshold. The sparse coefficients solved by this formula combined with the low-resolution dictionary can accurately represent low-resolution PET image blocks, but the solution of this formula is an NP-hard problem, and the formula can be equivalent to the process of solving the minimization of L1 normal form:

其中，λ为预设参数，用来平衡α的稀疏性和α与y的保真度。为了使得稀疏系数α的求解更为精确，即使得通过稀疏系数α恢复得到的高分辨率PET图像块与对应的低分辨率PET图像块的相关度更高，在这里对公式进行了更为准确的计算： Among them, λ is a preset parameter, which is used to balance the sparsity of α and the fidelity of α and y. In order to make the solution of the sparse coefficient α more accurate, that is, to make the high-resolution PET image block restored by the sparse coefficient α have a higher correlation with the corresponding low-resolution PET image block, the formula A more accurate calculation is made:

min||α||1，其中，α要求满足第一系数约束条件和第二系数约束条件其中，ε1为预设的第一阈值，ε2为预设的第二阈值，P用来提取当前低分辨率PET图像块与上一低分辨率PET图像块的重叠区域，ω为上一低分辨率PET图像块在重叠区域的值。最终，计算得到每块低分辨率PET图像块的稀疏系数。min||α|| 1 , where α is required to satisfy the first coefficient constraint and the second coefficient constraint Among them, ε1 is the preset first threshold, ε2 is the preset second threshold, P is used to extract the overlapping area between the current low-resolution PET image block and the previous low-resolution PET image block, and ω is the previous low-resolution PET image block. Values of low-resolution PET image patches in overlapping regions. Finally, the sparse coefficient of each low-resolution PET image block is calculated.

在步骤S205中，根据每个低分辨率PET图像块的稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像。In step S205, according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary, a high-resolution PET image corresponding to the low-resolution PET image is generated.

在本发明实施例中，每块低分辨率PET图像块对应的高分辨率PET图像可通过公式x＝D2α计算得到，其中，x为高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image corresponding to each low-resolution PET image block can be calculated by the formula x=D 2 α, where x is the high-resolution PET image block.

在步骤S206中，根据高分辨率PET图像，初始化低分辨率PET图像的第一重建图像。In step S206, the first reconstructed image of the low-resolution PET image is initialized according to the high-resolution PET image.

在本发明实施例中，由于在求解低分辨率PET图像块的稀疏系数时采用了近似逼近的方式，可能使得稀疏系数的准确度较低，求得的高分辨率PET图像质量不佳，因此生成并输出低分辨率PET图像的重建图像时需要对高分辨率PET图像进行优化。为了便于描述，将高分辨率PET图像设置为低分辨率PET图像的第一重建图像。In the embodiment of the present invention, due to the use of an approximate approximation method when solving the sparse coefficients of low-resolution PET image blocks, the accuracy of the sparse coefficients may be low, and the obtained high-resolution PET image quality is not good, so Generating and exporting reconstructions of low-resolution PET images requires optimization of high-resolution PET images. For the convenience of description, the high-resolution PET image is set as the first reconstructed image of the low-resolution PET image.

在步骤S207中，根据低分辨率PET图像、高分辨率PET图像、模糊矩阵、下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成低分辨率PET图像的第二重建图像。In step S207, optimize the first reconstructed image according to the low-resolution PET image, high-resolution PET image, blur matrix, downsampling matrix and preset gradient descent method, and generate the second reconstruction of the low-resolution PET image image.

在本发明实施例中，根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成第二重建图像，其中，梯度下降方式可表示为：In the embodiment of the present invention, the first reconstructed image is optimized according to the low-resolution PET image, high-resolution PET image, preset blur matrix, preset down-sampling matrix, and preset gradient descent method to generate the second Two reconstructed images, where the gradient descent method can be expressed as:

Xt+1＝Xt+v[HTST(Y-SHXt)+c(Xt-X0)]，其中，Xt为第t次优化过程中的第一重建图像，Xt+1为第t次优化过程中的第二重建图像，v为预设的梯度下降步长，H为模糊矩阵，S为矢量矩阵，Y为低分辨率PET图像，X0为高分辨率PET图像，c为预设参数。X t+1 =X t +v[H T S T (Y-SHX t )+c(X t -X 0 )], where X t is the first reconstructed image in the t-th optimization process, and X t +1 is the second reconstructed image in the t-th optimization process, v is the preset gradient descent step size, H is the blur matrix, S is the vector matrix, Y is the low-resolution PET image, X 0 is the high-resolution PET image, c is a preset parameter.

在步骤S208中，判断当前优化次数是否达到预设的最大优化次数。In step S208, it is determined whether the current number of optimization times reaches a preset maximum number of optimization times.

在本发明实施例中，判断当前优化次数是否达到预设的最大优化次数，当达到时，执行步骤S209，否则执行步骤S2010。In the embodiment of the present invention, it is judged whether the current number of optimization times reaches the preset maximum number of optimization times, and if it is reached, step S209 is executed; otherwise, step S2010 is executed.

在步骤S209中，输出第二重建图像。In step S209, a second reconstructed image is output.

在本发明实施例中，当当前优化次数达到最大优化次数时，可认为第二重建图像优化质量较好，可将第二重建图像设置为低分辨率PET图像的最终重建图像输出。In the embodiment of the present invention, when the current number of optimizations reaches the maximum number of optimizations, it can be considered that the optimization quality of the second reconstructed image is better, and the second reconstructed image can be set as the final reconstructed image output of the low-resolution PET image.

在步骤S210中，将第二重建图像设置为第一重建图像，对当前优化次数进行加一操作。In step S210, the second reconstructed image is set as the first reconstructed image, and an operation of adding one to the current number of optimizations is performed.

在本发明实施例中，当当前优化次数未达到最大优化次数时，可认为还需对第二重建图像进行优化，将第二重建图像设置为第一重建图像，将当前优化次数t进行加一操作，并跳转至执行根据低分辨率PET图像、高分辨率PET图像、模糊矩阵、下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化的步骤。In the embodiment of the present invention, when the current number of optimizations does not reach the maximum number of optimizations, it can be considered that the second reconstructed image needs to be optimized, the second reconstructed image is set as the first reconstructed image, and the current number of optimizations t is increased by one operation, and jump to the step of optimizing the first reconstructed image according to the low-resolution PET image, the high-resolution PET image, the blur matrix, the down-sampling matrix, and the preset gradient descent method.

在本发明实施例中，结合训练好的低分辨率字典和高分辨率字典构造第一系数约束条件和第二系数约束条件，根据第一系数约束条件和第二系数约束条件，对分割后的低分辨率PET图像块对应的稀疏系数进行近似计算，有效地提高低分辨率PET图像稀疏系数的准确度，根据稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，并通过梯度下降方式对高分辨率PET图像进行优化，有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建的计算复杂度，从而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the first coefficient constraint condition and the second coefficient constraint condition are constructed in combination with the trained low-resolution dictionary and high-resolution dictionary, and according to the first coefficient constraint condition and the second coefficient constraint condition, the divided The sparse coefficient corresponding to the low-resolution PET image block is approximated to effectively improve the accuracy of the low-resolution PET image sparse coefficient, and the high-resolution PET image corresponding to the low-resolution PET image is calculated according to the sparse coefficient and the high-resolution dictionary. And the high-resolution PET image is optimized by gradient descent, which effectively improves the image quality of PET image reconstruction, effectively reduces the computational complexity of PET image reconstruction, and thus effectively improves the reconstruction efficiency of PET image.

实施例三：Embodiment three:

图3示出了本发明实施例三提供的PET图像的重建装置的结构，为了便于说明，仅示出了与本发明实施例相关的部分，其中包括：Fig. 3 shows the structure of the PET image reconstruction device provided by the third embodiment of the present invention. For the convenience of description, only the parts related to the embodiment of the present invention are shown, including:

图像分割单元31，用于接收用户输入的低分辨率PET图像，对低分辨率PET图像进行分割，生成低分辨率PET图像块。The image segmentation unit 31 is configured to receive a low-resolution PET image input by a user, segment the low-resolution PET image, and generate low-resolution PET image blocks.

在本发明实施例中，可通过预设的图像分割算法对接收到的用户输入的低分辨率PET图像进行图像块的分割，生成多个低分辨率PET图像块，具体地，在分割时每个低分辨率PET图像块的尺寸相同，且前后相邻的低分辨率PET图像块之间存在重叠区域。In the embodiment of the present invention, the low-resolution PET image received by the user may be divided into image blocks by a preset image segmentation algorithm to generate multiple low-resolution PET image blocks. Specifically, each The size of the two low-resolution PET image blocks is the same, and there is an overlapping area between the adjacent low-resolution PET image blocks.

系数生成单元32，用于根据每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成每个低分辨率PET图像块的稀疏系数。The coefficient generation unit 32 is configured to generate sparse coefficients for each low-resolution PET image block according to each low-resolution PET image block, a pre-trained low-resolution dictionary and a high-resolution dictionary.

在本发明实施例中，低分辨率PET图像块对应的高分辨率PET图像块近似为高分辨率字典与低分辨率PET图像块的稀疏系数的乘积，因此需对低分辨率PET图像块的稀疏系数进行求解，以进一步得到高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image block corresponding to the low-resolution PET image block is approximately the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block. Sparse coefficients are solved to further obtain high-resolution PET image blocks.

在本发明实施例中，低分辨率PET图像块的稀疏系数可通过下列公式求解：In the embodiment of the present invention, the sparse coefficient of the low-resolution PET image block can be solved by the following formula:

min||α||0，且其中，F为预设的特征提取函数，α为低分辨率PET图像块y的稀疏系数，D1为低分辨率字典，ε为预设阈值。该公式求解出的稀疏系数结合低分辨率字典可精确地表示出低分辨率PET图像块，但该公式的求解是个NP-hard问题，可将该公式等效为求解L1范式最小化的过程：min||α|| 0 , and Among them, F is the preset feature extraction function, α is the sparse coefficient of the low-resolution PET image block y, D 1 is the low-resolution dictionary, and ε is the preset threshold. The sparse coefficients solved by this formula combined with the low-resolution dictionary can accurately represent low-resolution PET image blocks, but the solution of this formula is an NP-hard problem, and the formula can be equivalent to the process of solving the minimization of L1 normal form:

其中，λ为预设参数，用来平衡α的稀疏性和α与y的保真度。为了使得稀疏系数α的求解更为精确，即使得通过稀疏系数α恢复得到的高分辨率PET图像块与对应的低分辨率PET图像块的相关度更高，我们对公式进行了更为准确的计算： Among them, λ is a preset parameter, which is used to balance the sparsity of α and the fidelity of α and y. In order to make the solution of the sparse coefficient α more accurate, that is, to make the high-resolution PET image block restored by the sparse coefficient α have a higher correlation with the corresponding low-resolution PET image block, we formulate A more accurate calculation is made:

min||α||1，其中，α要求满足第一系数约束条件和第二系数约束条件其中，ε1为预设的第一阈值，ε2为预设的第二阈值，P用来提取当前低分辨率PET图像块与上一低分辨率PET图像块的重叠区域，ω为上一低分辨率PET图像块在重叠区域的值。最终，计算得到每块低分辨率PET图像块的稀疏系数。min||α|| 1 , where α is required to satisfy the first coefficient constraint and the second coefficient constraint Among them, ε1 is the preset first threshold, ε2 is the preset second threshold, P is used to extract the overlapping area between the current low-resolution PET image block and the previous low-resolution PET image block, and ω is the previous low-resolution PET image block. Values of low-resolution PET image patches in overlapping regions. Finally, the sparse coefficient of each low-resolution PET image block is calculated.

图像生成单元33，用于根据每个低分辨率PET图像块的稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像。The image generation unit 33 is configured to generate a high-resolution PET image corresponding to the low-resolution PET image according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary.

在本发明实施例中，每块低分辨率PET图像块对应的高分辨率PET图像可通过公式x＝D2α计算得到，其中，x为高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image corresponding to each low-resolution PET image block can be calculated by the formula x=D 2 α, where x is the high-resolution PET image block.

图像输出单元34，用于根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵和预设的下采样矩阵，生成并输出低分辨率PET图像的重建图像。The image output unit 34 is configured to generate and output a reconstructed image of the low-resolution PET image according to the low-resolution PET image, the high-resolution PET image, a preset blur matrix and a preset downsampling matrix.

在本发明实施例中，由于在求解低分辨率PET图像块的稀疏系数时采用了近似逼近的方式，可能使得稀疏系数的准确度较低，求得的高分辨率PET图像质量不佳，因此需要对高分辨率PET图像进行优化。为了便于描述，将高分辨率PET图像设置为低分辨率PET图像的第一重建图像，根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成第二重建图像，其中，梯度下降方式可表示为：In the embodiment of the present invention, due to the use of an approximate approximation method when solving the sparse coefficients of low-resolution PET image blocks, the accuracy of the sparse coefficients may be low, and the obtained high-resolution PET image quality is not good, so Optimization is required for high-resolution PET images. For the convenience of description, the high-resolution PET image is set as the first reconstructed image of the low-resolution PET image, according to the low-resolution PET image, high-resolution PET image, preset blur matrix, preset downsampling matrix and preset The gradient descent method is set to optimize the first reconstructed image to generate the second reconstructed image, where the gradient descent method can be expressed as:

Xt+1＝Xt+v[HTST(Y-SHXt)+c(Xt-X0)]，其中，Xt为第t次优化过程中的第一重建图像，Xt+1为第t次优化过程中的第二重建图像，v为预设的梯度下降步长，H为模糊矩阵，S为矢量矩阵，Y为低分辨率PET图像，X0为高分辨率PET图像，c为预设参数。判断当前优化次数t是否达到预设的最大优化次数，当达到时，输出第二重建图像，否则，将第二重建图像设置为第一重建图像，对当前优化次数t进行加一操作，并跳转至执行根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化的步骤，从而通过对高分辨率PET图像进行优化，有效地提高了低分辨率PET图像重建的质量，并有效地降低了低分辨率PET图像重建的计算复杂度。X t+1 =X t +v[H T S T (Y-SHX t )+c(X t -X 0 )], where X t is the first reconstructed image in the t-th optimization process, and X t +1 is the second reconstructed image in the t-th optimization process, v is the preset gradient descent step size, H is the blur matrix, S is the vector matrix, Y is the low-resolution PET image, X 0 is the high-resolution PET image, c is a preset parameter. Judging whether the current optimization times t has reached the preset maximum number of optimization times, when reached, output the second reconstructed image, otherwise, set the second reconstructed image as the first reconstructed image, add one to the current optimization times t, and skip Go to the step of optimizing the first reconstructed image according to the low-resolution PET image, the high-resolution PET image, the preset blur matrix, the preset down-sampling matrix and the preset gradient descent method, so as to pass the high-resolution The optimization of high-resolution PET images effectively improves the quality of low-resolution PET image reconstruction and effectively reduces the computational complexity of low-resolution PET image reconstruction.

在本发明实施例中，结合训练好的低分辨率字典和高分辨率字典构造第一系数约束条件和第二系数约束条件，根据第一系数约束条件和第二系数约束条件，对分割后的低分辨率PET图像块对应的稀疏系数进行近似计算，有效地提高低分辨率PET图像稀疏系数的准确度，根据稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，并通过梯度下降方式对高分辨率PET图像进行优化，有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建的计算复杂度，从而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the first coefficient constraint condition and the second coefficient constraint condition are constructed in combination with the trained low-resolution dictionary and high-resolution dictionary, and according to the first coefficient constraint condition and the second coefficient constraint condition, the divided The sparse coefficient corresponding to the low-resolution PET image block is approximated to effectively improve the accuracy of the low-resolution PET image sparse coefficient, and the high-resolution PET image corresponding to the low-resolution PET image is calculated according to the sparse coefficient and the high-resolution dictionary. And the high-resolution PET image is optimized by gradient descent, which effectively improves the image quality of PET image reconstruction, effectively reduces the computational complexity of PET image reconstruction, and thus effectively improves the reconstruction efficiency of PET image.

实施例四：Embodiment four:

图4示出了本发明实施例四提供的PET图像的重建装置的结构，为了便于说明，仅示出了与本发明实施例相关的部分，其中包括：Figure 4 shows the structure of the PET image reconstruction device provided by Embodiment 4 of the present invention. For the convenience of description, only the parts related to the embodiment of the present invention are shown, including:

字典初始化单元41，用于对低分辨率字典和高分辨率字典进行随机初始化。The dictionary initialization unit 41 is configured to randomly initialize the low-resolution dictionary and the high-resolution dictionary.

字典训练单元42，用于根据预设的低分辨率PET训练图集、预设的高分辨率PET训练图集、低分辨率PET训练图集中图像块的尺寸、以及高分辨率PET训练图集中图像块的尺寸，对低分辨率字典和高分辨率字典进行联合训练。The dictionary training unit 42 is configured to use the preset low-resolution PET training atlas, the preset high-resolution PET training atlas, the size of the image blocks in the low-resolution PET training atlas, and the high-resolution PET training atlas. Dimensions of image patches for joint training of low-resolution and high-resolution dictionaries.

在本发明实施例中，可通过高斯随机矩阵对低分辨率字典和高分辨率字典进行随机初始化，由于高分辨字典与低分辨率PET图像块的稀疏系数的乘积可近似为高分辨率PET图像块，而低分辨率PET图像块的稀疏系数与低分辨率字典的乘积为低分辨率PET图像块，可看出高分辨率字典与低分辨率字典间具有联系，因此需对低分辨率字典和高分辨率字典进行联合训练。具体地，联合训练的公式为：In the embodiment of the present invention, the low-resolution dictionary and the high-resolution dictionary can be randomly initialized by a Gaussian random matrix, because the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block can be approximated as a high-resolution PET image block, and the product of the sparse coefficient of the low-resolution PET image block and the low-resolution dictionary is the low-resolution PET image block. It can be seen that there is a connection between the high-resolution dictionary and the low-resolution dictionary, so the low-resolution dictionary Joint training with high-resolution dictionaries. Specifically, the formula for joint training is:

其中，c＝1,2，X1为预设的低分辨率PET训练图集中低分辨率PET训练图像，X2为预设的高分辨率PET训练图集中高分辨率PET训练图像，Z为预设的矩阵变量，N为低分辨率PET训练图像的图像块尺寸，M为高分辨率PET训练图像的图像块尺寸。 Among them, c=1,2, X 1 is the low-resolution PET training image in the preset low-resolution PET training atlas, X 2 is the high-resolution PET training image in the preset high-resolution PET training atlas, and Z is Preset matrix variables, N is the image block size of the low-resolution PET training image, and M is the image block size of the high-resolution PET training image.

图像分割单元43，用于接收用户输入的低分辨率PET图像，对低分辨率PET图像进行分割，生成低分辨率PET图像块。The image segmentation unit 43 is configured to receive the low-resolution PET image input by the user, segment the low-resolution PET image, and generate low-resolution PET image blocks.

在本发明实施例中，可通过预设的图像分割算法对接收到的用户输入的低分辨率PET图像进行图像块的分割，生成低分辨率PET图像块，每个低分辨率PET图像块的尺寸相同，前后相邻的低分辨率PET图像块之间存在重叠区域。In the embodiment of the present invention, the low-resolution PET image received by the user may be segmented into image blocks by a preset image segmentation algorithm to generate low-resolution PET image blocks, and each low-resolution PET image block The size is the same, and there is an overlapping area between the adjacent low-resolution PET image blocks.

系数生成单元44，用于根据每个低分辨率PET图像块、预先训练好的低分辨率字典和高分辨率字典，生成每个低分辨率PET图像块的稀疏系数。The coefficient generation unit 44 is configured to generate sparse coefficients for each low-resolution PET image block according to each low-resolution PET image block, a pre-trained low-resolution dictionary and a high-resolution dictionary.

在本发明实施例中，低分辨率PET图像块对应的高分辨率PET图像块近似为高分辨率字典与低分辨率PET图像块的稀疏系数的乘积，因此需对低分辨率PET图像块的稀疏系数进行求解，以进一步得到高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image block corresponding to the low-resolution PET image block is approximately the product of the high-resolution dictionary and the sparse coefficient of the low-resolution PET image block. Sparse coefficients are solved to further obtain high-resolution PET image blocks.

在本发明实施例中，低分辨率PET图像块的稀疏系数可通过下列公式求解：In the embodiment of the present invention, the sparse coefficient of the low-resolution PET image block can be solved by the following formula:

min||α||0，且其中，F为预设的特征提取函数，α为低分辨率PET图像块y的稀疏系数，D1为低分辨率字典，ε为预设阈值。该公式求解出的稀疏系数结合低分辨率字典可精确地表示出低分辨率PET图像块，但该公式的求解是个NP-hard问题，可将该公式等效为求解L1范式最小化的过程：min||α|| 0 , and Among them, F is the preset feature extraction function, α is the sparse coefficient of the low-resolution PET image block y, D 1 is the low-resolution dictionary, and ε is the preset threshold. The sparse coefficients solved by this formula combined with the low-resolution dictionary can accurately represent low-resolution PET image blocks, but the solution of this formula is an NP-hard problem, and the formula can be equivalent to the process of solving the minimization of L1 normal form:

其中，λ为预设参数，用来平衡α的稀疏性和α与y的保真度。为了使得稀疏系数α的求解更为精确，即使得通过稀疏系数α恢复得到的高分辨率PET图像块与对应的低分辨率PET图像块的相关度更高，在这里对公式进行了更为准确的计算： Among them, λ is a preset parameter, which is used to balance the sparsity of α and the fidelity of α and y. In order to make the solution of the sparse coefficient α more accurate, that is, to make the high-resolution PET image block restored by the sparse coefficient α have a higher correlation with the corresponding low-resolution PET image block, the formula A more accurate calculation is made:

min||α||1，其中，α要求满足第一系数约束条件和第二系数约束条件其中，ε1为预设的第一阈值，ε2为预设的第二阈值，P用来提取当前低分辨率PET图像块与上一低分辨率PET图像块的重叠区域，ω为上一低分辨率PET图像块在重叠区域的值。最终，计算得到每块低分辨率PET图像块的稀疏系数。min||α|| 1 , where α is required to satisfy the first coefficient constraint and the second coefficient constraint Among them, ε1 is the preset first threshold, ε2 is the preset second threshold, P is used to extract the overlapping area between the current low-resolution PET image block and the previous low-resolution PET image block, and ω is the previous low-resolution PET image block. Values of low-resolution PET image patches in overlapping regions. Finally, the sparse coefficient of each low-resolution PET image block is calculated.

图像生成单元45，用于根据每个低分辨率PET图像块的稀疏系数和高分辨率字典，生成低分辨率PET图像对应的高分辨率PET图像。The image generation unit 45 is configured to generate a high-resolution PET image corresponding to the low-resolution PET image according to the sparse coefficients of each low-resolution PET image block and the high-resolution dictionary.

在本发明实施例中，每块低分辨率PET图像块对应的高分辨率PET图像可通过公式x＝D2α计算得到，其中，x为高分辨率PET图像块。In the embodiment of the present invention, the high-resolution PET image corresponding to each low-resolution PET image block can be calculated by the formula x=D 2 α, where x is the high-resolution PET image block.

图像输出单元46，用于根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵和预设的下采样矩阵，生成并输出低分辨率PET图像的重建图像。The image output unit 46 is configured to generate and output a reconstructed image of the low-resolution PET image according to the low-resolution PET image, the high-resolution PET image, a preset blur matrix and a preset downsampling matrix.

在本发明实施例中，由于在求解低分辨率PET图像块的稀疏系数时采用了近似逼近的方式，可能使得稀疏系数的准确度较低，求得的高分辨率PET图像质量不佳，因此生成并输出低分辨率PET图像的重建图像时需要对高分辨率PET图像进行优化。为了便于描述，将高分辨率PET图像设置为低分辨率PET图像的第一重建图像，根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成第二重建图像，其中，梯度下降方式可表示为：In the embodiment of the present invention, due to the use of an approximate approximation method when solving the sparse coefficients of low-resolution PET image blocks, the accuracy of the sparse coefficients may be low, and the obtained high-resolution PET image quality is not good, so Generating and exporting reconstructions of low-resolution PET images requires optimization of high-resolution PET images. For the convenience of description, the high-resolution PET image is set as the first reconstructed image of the low-resolution PET image, according to the low-resolution PET image, high-resolution PET image, preset blur matrix, preset downsampling matrix and preset The gradient descent method is set to optimize the first reconstructed image to generate the second reconstructed image, where the gradient descent method can be expressed as:

Xt+1＝Xt+v[HTST(Y-SHXt)+c(Xt-X0)]，其中，Xt为第t次优化过程中的第一重建图像，Xt+1为第t次优化过程中的第二重建图像，v为预设的梯度下降步长，H为模糊矩阵，S为矢量矩阵，Y为低分辨率PET图像，X0为高分辨率PET图像，c为预设参数。判断当前优化次数t是否达到预设的最大优化次数，当达到时，输出第二重建图像，否则，将第二重建图像设置为第一重建图像，对当前优化次数t进行加一操作，并跳转至执行根据低分辨率PET图像、高分辨率PET图像、预设的模糊矩阵、预设的下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化的步骤，从而通过对高分辨率PET图像进行优化，有效地提高了低分辨率PET图像重建的质量，并有效地降低了低分辨率PET图像重建的计算复杂度。X t+1 =X t +v[H T S T (Y-SHX t )+c(X t -X 0 )], where X t is the first reconstructed image in the t-th optimization process, and X t +1 is the second reconstructed image in the t-th optimization process, v is the preset gradient descent step size, H is the blur matrix, S is the vector matrix, Y is the low-resolution PET image, X 0 is the high-resolution PET image, c is a preset parameter. Judging whether the current optimization times t has reached the preset maximum number of optimization times, when reached, output the second reconstructed image, otherwise, set the second reconstructed image as the first reconstructed image, add one to the current optimization times t, and skip Go to the step of optimizing the first reconstructed image according to the low-resolution PET image, the high-resolution PET image, the preset blur matrix, the preset down-sampling matrix and the preset gradient descent method, so as to pass the high-resolution The optimization of high-resolution PET images effectively improves the quality of low-resolution PET image reconstruction and effectively reduces the computational complexity of low-resolution PET image reconstruction.

优选地，系数生成单元44包括第一约束构建单元441、第二约束构建单元442和系数计算单元443，其中：Preferably, the coefficient generation unit 44 includes a first constraint construction unit 441, a second constraint construction unit 442 and a coefficient calculation unit 443, wherein:

第一约束构建单元441，用于根据低分辨率PET图像块、低分辨率字典、预设的特征提取函数和预设第一阈值，构建第一系数约束条件；The first constraint construction unit 441 is configured to construct a first coefficient constraint condition according to the low-resolution PET image block, the low-resolution dictionary, the preset feature extraction function and the preset first threshold;

第二约束构建单元442，用于根据低分辨率PET图像块与低分辨率PET图像块的上一个低分辨率PET图像块的重叠区域、高分辨率字典和预设的第二阈值，构建第二系数约束条件；以及The second constraint construction unit 442 is configured to construct the second constraint according to the overlapping area between the low-resolution PET image block and the previous low-resolution PET image block of the low-resolution PET image block, the high-resolution dictionary and the preset second threshold. Two coefficient constraints; and

系数计算单元443，用于根据预设的系数计算公式，计算满足第一系数约束条件和第二系数约束条件的、低分辨率PET图像块的稀疏系数。The coefficient calculation unit 443 is configured to calculate the sparse coefficients of the low-resolution PET image block satisfying the first coefficient constraint condition and the second coefficient constraint condition according to a preset coefficient calculation formula.

优选地，图像输出单元46包括重建图像初始化单元461、重建图像优化单元462、重建判断单元463和重建输出单元464，其中：Preferably, the image output unit 46 includes a reconstructed image initialization unit 461, a reconstructed image optimization unit 462, a reconstruction judgment unit 463 and a reconstruction output unit 464, wherein:

重建图像初始化单元461，用于根据高分辨率PET图像，初始化低分辨率PET图像的第一重建图像；A reconstructed image initialization unit 461, configured to initialize the first reconstructed image of the low-resolution PET image according to the high-resolution PET image;

重建图像优化单元462，用于根据低分辨率PET图像、高分辨率PET图像、模糊矩阵、下采样矩阵和预设的梯度下降方式，对第一重建图像进行优化，生成低分辨率PET图像的第二重建图像；The reconstructed image optimization unit 462 is configured to optimize the first reconstructed image according to the low-resolution PET image, the high-resolution PET image, the blur matrix, the down-sampling matrix and the preset gradient descent method, and generate the low-resolution PET image a second reconstructed image;

重建判断单元463，用于判断当前优化次数是否达到预设的最大优化次数；以及A reconstruction judging unit 463, configured to judge whether the current number of optimizations reaches a preset maximum number of optimizations; and

重建输出单元464，用于当当前优化次数达到最大优化次数时，输出第二重建图像，否则，将第二重建图像设置为第一重建图像，对当前优化次数进行加一操作，并由重建图像优化单元462执行对第一重建图像进行优化的操作。The reconstruction output unit 464 is used to output the second reconstructed image when the current optimization times reach the maximum optimization times, otherwise, set the second reconstructed image as the first reconstructed image, add one to the current optimization times, and reconstruct the image The optimization unit 462 performs an operation of optimizing the first reconstructed image.

在本发明实施例中，结合训练好的低分辨率字典和高分辨率字典构造第一系数约束条件和第二系数约束条件，根据第一系数约束条件和第二系数约束条件，对分割后的低分辨率PET图像块对应的稀疏系数进行近似计算，有效地提高低分辨率PET图像稀疏系数的准确度，根据稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，并通过梯度下降方式对高分辨率PET图像进行优化，有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建的计算复杂度，从而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the first coefficient constraint condition and the second coefficient constraint condition are constructed in combination with the trained low-resolution dictionary and high-resolution dictionary, and according to the first coefficient constraint condition and the second coefficient constraint condition, the divided The sparse coefficient corresponding to the low-resolution PET image block is approximated to effectively improve the accuracy of the low-resolution PET image sparse coefficient, and the high-resolution PET image corresponding to the low-resolution PET image is calculated according to the sparse coefficient and the high-resolution dictionary. And the high-resolution PET image is optimized by gradient descent, which effectively improves the image quality of PET image reconstruction, effectively reduces the computational complexity of PET image reconstruction, and thus effectively improves the reconstruction efficiency of PET image.

在本发明实施例中，图像重建装置的各单元可由相应的硬件或软件单元实现，各单元可以为独立的软、硬件单元，也可以集成为一个软、硬件单元，在此不用以限制本发明。In the embodiment of the present invention, each unit of the image reconstruction device can be realized by a corresponding hardware or software unit, and each unit can be an independent software and hardware unit, or can be integrated into a software and hardware unit, which is not intended to limit the present invention .

实施例五：Embodiment five:

图5示出了本发明实施例五提供的医学图像处理设备的结构，为了便于说明，仅示出了与本发明实施例相关的部分。FIG. 5 shows the structure of a medical image processing device provided by Embodiment 5 of the present invention. For convenience of description, only parts related to the embodiment of the present invention are shown.

本发明实施例的医学图像处理设备5包括处理器50、存储器51以及存储在存储器51中并可在处理器50上运行的计算机程序52。该处理器50执行计算机程序52时实现上述各个方法实施例中的步骤，例如图1所示的步骤S101至S104。或者，处理器50执行计算机程序52时实现上述各装置实施例中各单元的功能，例如图3所示单元31至34的功能。The medical image processing device 5 of the embodiment of the present invention includes a processor 50 , a memory 51 and a computer program 52 stored in the memory 51 and operable on the processor 50 . When the processor 50 executes the computer program 52, the steps in the above-mentioned method embodiments are implemented, for example, steps S101 to S104 shown in FIG. 1 . Alternatively, when the processor 50 executes the computer program 52, the functions of the units in the above-mentioned device embodiments are implemented, for example, the functions of the units 31 to 34 shown in FIG. 3 .

在本发明实施例中，根据分割后的低分辨率PET图像块、训练好的低分辨率字典和高分辨率字典，计算每个低分辨率PET图像块的稀疏系数，根据这些稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，通过对高分辨率PET图像进行后处理，得到低分辨率PET的重建图像，从而有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建中的计算复杂度，进而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the sparse coefficient of each low-resolution PET image block is calculated according to the segmented low-resolution PET image block, the trained low-resolution dictionary and the high-resolution dictionary, and according to these sparse coefficients and high The resolution dictionary calculates the high-resolution PET image corresponding to the low-resolution PET image, and obtains the reconstructed image of the low-resolution PET image by post-processing the high-resolution PET image, thereby effectively improving the image quality of the PET image reconstruction, effectively The computational complexity in PET image reconstruction is greatly reduced, and the reconstruction efficiency of PET images is effectively improved.

实施例六：Embodiment six:

在本发明实施例中，提供了一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序，该计算机程序被处理器执行时实现上述各个方法实施例中的步骤，例如，图1所示的步骤S101至S104。或者，该计算机程序被处理器执行时实现上述各装置实施例中各单元的功能，例如图3所示单元31至34的功能。In an embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented. For example, FIG. 1 Steps S101 to S104 are shown. Alternatively, when the computer program is executed by the processor, the functions of the units in the above-mentioned device embodiments are realized, for example, the functions of the units 31 to 34 shown in FIG. 3 .

在本发明实施例中，根据分割后的低分辨率PET图像块、训练好的低分辨率字典和高分辨率字典，计算每个低分辨率PET图像块的稀疏系数，根据这些稀疏系数和高分辨率字典计算低分辨率PET图像对应的高分辨率PET图像，通过对高分辨率PET图像进行后处理，得到低分辨率PET的重建图像，从而有效地提高了PET图像重建的图像质量，有效地降低了PET图像重建中的计算复杂度，进而有效地提高了PET图像的重建效率。In the embodiment of the present invention, the sparse coefficient of each low-resolution PET image block is calculated according to the segmented low-resolution PET image block, the trained low-resolution dictionary and the high-resolution dictionary, and according to these sparse coefficients and high The resolution dictionary calculates the high-resolution PET image corresponding to the low-resolution PET image, and obtains the reconstructed image of the low-resolution PET image by post-processing the high-resolution PET image, thereby effectively improving the image quality of the PET image reconstruction, effectively The computational complexity in PET image reconstruction is greatly reduced, and the reconstruction efficiency of PET images is effectively improved.

本发明实施例的计算机可读存储介质可以包括能够携带计算机程序代码的任何实体或装置、记录介质，例如，ROM/RAM、磁盘、光盘、闪存等存储器。The computer-readable storage medium in the embodiments of the present invention may include any entity or device or recording medium capable of carrying computer program codes, such as ROM/RAM, magnetic disk, optical disk, flash memory and other memories.

以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.

Claims (10)

1. a kind of method for reconstructing of PET image, it is characterised in that methods described comprises the steps：
The low resolution PET image of user's input is received, the low resolution PET image is split, generates low resolution PET image block；
According to the good low-resolution dictionary of each low resolution PET image block, training in advance and high-resolution dictionary, generation The sparse coefficient of each low resolution PET image block；
According to the sparse coefficient of each low resolution PET image block and the high-resolution dictionary, the low resolution is generated High-resolution PET image corresponding to rate PET image；
According to the low resolution PET image, the high-resolution PET image, default fuzzy matrix and default down-sampling Matrix, generate and export the reconstruction image of the low resolution PET image.
2. the method as described in claim 1, it is characterised in that according to each low resolution PET image block, training in advance Good low-resolution dictionary and high-resolution dictionary, the step of generating the sparse coefficient of each low resolution PET image block, Including：
According to the low resolution PET image block, the low-resolution dictionary, default feature extraction function and default first threshold Value, build the first restricted coefficients of equation condition；
According to the low resolution PET image block and the upper low resolution PET image block of the low resolution PET image block Overlapping region, the high-resolution dictionary and default Second Threshold, build the second restricted coefficients of equation condition；
According to default coefficient formulas, calculating meets the first restricted coefficients of equation condition and the second restricted coefficients of equation condition , the sparse coefficient of the low resolution PET image block.
3. the method as described in claim 1, it is characterised in that according to the low resolution PET image, the high-resolution PET image, default fuzzy matrix and default down-sampling matrix, generate and export the reconstruction of the low resolution PET image The step of image, including：
According to the high-resolution PET image, the first reconstruction image of the low resolution PET image is initialized；
According to the low resolution PET image, the high-resolution PET image, the fuzzy matrix, the down-sampling matrix and Default gradient declines mode, and first reconstruction image is optimized, generates the second weight of the low resolution PET image Build image；
Judge whether current optimization number reaches default largest optimization number；
When the current optimization number reaches the largest optimization number, second reconstruction image is exported, otherwise, by described in Second reconstruction image is arranged to first reconstruction image, and the current optimization number is carried out plus one operates, and jumps to and holds The step of row optimizes to first reconstruction image.
4. the method as described in claim 1, it is characterised in that receive user input low resolution PET image the step of it Before, methods described also includes：
Random initializtion is carried out to the low-resolution dictionary and the high-resolution dictionary；
Atlas, default high-resolution PET training atlas, low resolution PET instructions are trained according to default low resolution PET Practice the size that the size of image block and the high-resolution PET in atlas train image block in atlas, to the low resolution Dictionary and the high-resolution dictionary carry out joint training.
5. a kind of reconstructing device of PET image, it is characterised in that described device includes：
Image segmentation unit, for receiving the low resolution PET image of user's input, the low resolution PET image is carried out Segmentation, generate low resolution PET image block；
Coefficient generation unit, for according to the good low-resolution dictionary of each low resolution PET image block, training in advance and High-resolution dictionary, generate the sparse coefficient of each low resolution PET image block；
Image generation unit, for the sparse coefficient according to each low resolution PET image block and the high-resolution word Allusion quotation, generate high-resolution PET image corresponding to the low resolution PET image；And
Image output unit, for according to the low resolution PET image, the high-resolution PET image, default fuzzy square Battle array and default down-sampling matrix, generate and export the reconstruction image of the low resolution PET image.
6. device as claimed in claim 5, it is characterised in that the coefficient generation unit includes：
First constraint construction unit, for according to the low resolution PET image block, the low-resolution dictionary, default spy Sign extraction function and preset first threshold value, build the first restricted coefficients of equation condition；
Second constraint construction unit, for according to the low resolution PET image block and the low resolution PET image block The overlapping region of one low resolution PET image block, the high-resolution dictionary and default Second Threshold, build the second coefficient Constraints；And
Coefficient calculation unit, for meeting the first restricted coefficients of equation condition and institute according to default coefficient formulas, calculating State the second restricted coefficients of equation condition, the low resolution PET image block sparse coefficient.
7. device as claimed in claim 5, it is characterised in that described image output unit includes：
Reconstruction image initialization unit, for according to the high-resolution PET image, initializing the low resolution PET image The first reconstruction image；
Reconstruction image optimizes unit, for according to the low resolution PET image, the high-resolution PET image, described fuzzy Matrix, the down-sampling matrix and default gradient decline mode, and first reconstruction image is optimized, and generation is described low Second reconstruction image of resolution PET images；
Judging unit is rebuild, for judging currently to optimize whether number reaches default largest optimization number；And
Output unit is rebuild, for when the current optimization number reaches the largest optimization number, exporting second weight Image is built, otherwise, second reconstruction image is arranged to first reconstruction image, the current optimization number is added One operation, and the operation optimized to first reconstruction image is performed by reconstruction image optimization unit.
8. device as claimed in claim 5, it is characterised in that described device also includes：
Dictionary initialization unit, for carrying out random initializtion to the low-resolution dictionary and the high-resolution dictionary；With And
Dictionary training unit, for according to default low resolution PET train atlas, default high-resolution PET train atlas, The size of image block and the high-resolution PET train the chi of image block in atlas in the low resolution PET training atlas It is very little, joint training is carried out to the low-resolution dictionary and the high-resolution dictionary.
9. a kind of Medical Image Processing equipment, including memory, processor and it is stored in the memory and can be described The computer program run on processor, it is characterised in that realize such as right described in the computing device during computer program It is required that the step of any one of 1 to 4 methods described.
10. a kind of computer-readable recording medium, the computer-readable recording medium storage has computer program, and its feature exists In when the computer program is executed by processor the step of realization such as any one of Claims 1-4 methods described.
CN201710821065.6A 2017-09-13 2017-09-13 Method for reconstructing, device, equipment and the storage medium of PET image Pending CN107680146A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710821065.6A CN107680146A (en) 2017-09-13 2017-09-13 Method for reconstructing, device, equipment and the storage medium of PET image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710821065.6A CN107680146A (en) 2017-09-13 2017-09-13 Method for reconstructing, device, equipment and the storage medium of PET image

Publications (1)

Publication Number Publication Date
CN107680146A true CN107680146A (en) 2018-02-09

Family

ID=61135897

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710821065.6A Pending CN107680146A (en) 2017-09-13 2017-09-13 Method for reconstructing, device, equipment and the storage medium of PET image

Country Status (1)

Country Link
CN (1) CN107680146A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109741255A (en) * 2018-12-12 2019-05-10 深圳先进技术研究院 PET image super-resolution reconstruction method, device, equipment and medium based on decision tree
WO2020118844A1 (en) * 2018-12-14 2020-06-18 深圳先进技术研究院 Reconstruction method for pet image, computer storage medium, and computer device

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105844590A (en) * 2016-03-23 2016-08-10 武汉理工大学 Image super-resolution reconstruction method and system based on sparse representation
CN106227015A (en) * 2016-07-11 2016-12-14 中国科学院深圳先进技术研究院 A kind of hologram image super-resolution reconstruction method and system based on compressive sensing theory
CN106491151A (en) * 2016-01-25 2017-03-15 上海联影医疗科技有限公司 PET image acquisition methods and system
CN106780342A (en) * 2016-12-28 2017-05-31 深圳市华星光电技术有限公司 Single-frame image super-resolution reconstruction method and device based on the reconstruct of sparse domain
CN106846430A (en) * 2014-11-21 2017-06-13 上海联影医疗科技有限公司 A kind of image rebuilding method
CN106934766A (en) * 2017-03-15 2017-07-07 西安理工大学 A kind of infrared image super resolution ratio reconstruction method based on rarefaction representation
CN106952317A (en) * 2017-03-23 2017-07-14 西安电子科技大学 Hyperspectral Image Reconstruction Method Based on Structure Sparse

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106846430A (en) * 2014-11-21 2017-06-13 上海联影医疗科技有限公司 A kind of image rebuilding method
CN106491151A (en) * 2016-01-25 2017-03-15 上海联影医疗科技有限公司 PET image acquisition methods and system
CN105844590A (en) * 2016-03-23 2016-08-10 武汉理工大学 Image super-resolution reconstruction method and system based on sparse representation
CN106227015A (en) * 2016-07-11 2016-12-14 中国科学院深圳先进技术研究院 A kind of hologram image super-resolution reconstruction method and system based on compressive sensing theory
CN106780342A (en) * 2016-12-28 2017-05-31 深圳市华星光电技术有限公司 Single-frame image super-resolution reconstruction method and device based on the reconstruct of sparse domain
CN106934766A (en) * 2017-03-15 2017-07-07 西安理工大学 A kind of infrared image super resolution ratio reconstruction method based on rarefaction representation
CN106952317A (en) * 2017-03-23 2017-07-14 西安电子科技大学 Hyperspectral Image Reconstruction Method Based on Structure Sparse

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭坦: "基于稀疏表示的超分辨率图像重建算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109741255A (en) * 2018-12-12 2019-05-10 深圳先进技术研究院 PET image super-resolution reconstruction method, device, equipment and medium based on decision tree
WO2020118844A1 (en) * 2018-12-14 2020-06-18 深圳先进技术研究院 Reconstruction method for pet image, computer storage medium, and computer device

Similar Documents

Publication Publication Date Title
US11501415B2 (en) Method and system for high-resolution image inpainting
WO2020168844A1 (en) Image processing method, apparatus, equipment, and storage medium
JP6998968B2 (en) Deep neural network execution method, execution device, learning method, learning device and program
CN112132959B (en) Digital rock core image processing method and device, computer equipment and storage medium
CN110827216A (en) Multi-generator generation countermeasure network learning method for image denoising
CN112396645B (en) Monocular image depth estimation method and system based on convolution residual learning
CN105469363B (en) Method and apparatus based on blurred picture generation clear image
Zhao et al. Unsupervised degradation learning for single image super-resolution
Ziabari et al. 2.5 D deep learning for CT image reconstruction using a multi-GPU implementation
CN113496466A (en) Systems and methods for training networks for real-world super-resolution
CN106127688B (en) A super-resolution image reconstruction method and system thereof
Rivadeneira et al. Thermal image super-resolution challenge-pbvs 2021
Ting et al. Image inpainting by global structure and texture propagation
Liu et al. Video frame interpolation via optical flow estimation with image inpainting
Sun et al. The tenth ntire 2025 image denoising challenge report
CN108038840B (en) Image processing method and device, image processing equipment and storage medium
CN118172543A (en) A method and system for detecting small targets in infrared images based on deep expansion of transform domain tensor
Li et al. 4× Super‐resolution of unsupervised CT images based on GAN
CN112017113B (en) Image processing method and device, model training method and device, equipment and medium
CN107680146A (en) Method for reconstructing, device, equipment and the storage medium of PET image
CN105590296A (en) Dual-dictionary learning-based single-frame image super-resolution reconstruction method
Liu et al. Facial image inpainting using multi-level generative network
Silva et al. Efficient algorithm for convolutional dictionary learning via accelerated proximal gradient consensus
Wu et al. Non‐uniform image blind deblurring by two‐stage fully convolution network
CN117576260A (en) Pose generation method based on diffusion model and ControlNet

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20180209