Multi-mode square core optical fiber imaging system

Abstract

The invention belongs to the technical field of optical fiber imaging, and particularly relates to a multimode square core optical fiber imaging system which comprises a spatial light beam expanding module, a DMD optical fiber imaging module, a multimode square core optical fiber, a control and acquisition module and a deep learning image recovery algorithm module. The spatial light beam expanding module comprises an optical fiber laser, an optical fiber collimating lens, a reflecting mirror, a first positive focal lens and a second positive focal lens; the reflecting mirror is positioned between the optical fiber collimating lens and the first positive focal lens and is obliquely distributed; the DMD optical fiber imaging module comprises a digital micromirror device, a first plano-convex lens, a focusing objective lens, a beam expanding objective lens, a second plano-convex lens and a CCD (Charge Coupled Device); and the digital micromirror device is positioned between the second positive focus lens and the first plano-convex lens. According to the invention, the problem of circular contour bottom noise in traditional multimode optical fiber imaging can be effectively solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of fiber optic imaging, and particularly relates to a multimode square-core fiber optic imaging system. Background Art

[0002] Due to the property of the core diameter of multimode fiber being dozens to hundreds of micrometers, it can support thousands of independently propagating modes and can transmit much higher information density than single-mode fiber. In recent years, the development of deep learning technology has provided a data-driven end-to-end multimode fiber image imaging technology, which has attracted much attention and favor from researchers in the field of multimode fiber image transmission and restoration.

[0003] In 2018, researchers in Switzerland first used the UNet network to achieve the restoration of the MNIST dataset (binary images) based on multimode fiber (Optica, 2018, 5(8): 960-966.). Since then, with the rapid development of deep learning, many researchers have applied neural networks with different structures and image restoration schemes with different optical paths to the reconstruction of multimode fiber images, achieving the restoration from binary images to grayscale images ([1] Laser & Photonics Reviews, 2022, 16(9): 2100724, [2] Lu Wenkai, Wang Yonghao. Multimode fiber imaging method based on Fourier transform artificial neural network [P]. Beijing: CN202110288779.1, 2023-07-18.). However, in these multimode fiber restoration schemes, traditional cylindrical multimode fibers are used, and there will be circular contour background noise in the restored image results, which greatly limits the application and development of multimode fiber image restoration. Summary of the Invention

[0004] The purpose of the present invention is to provide a multimode square-core fiber optic imaging system, which can effectively solve the problem of circular contour background noise in traditional multimode fiber imaging.

[0005] The technical solution adopted by the present invention is specifically as follows:

[0006] A multimode square-core fiber optic imaging system includes a spatial light beam expander module, a DMD fiber imaging module, a multimode square-core fiber, a control and acquisition module, and a deep learning image restoration algorithm module;

[0007] The spatial light beam expander module includes a fiber laser, a fiber collimator, a reflector, a first positive focal length lens, and a second positive focal length lens. The reflector is located between the fiber collimator and the first positive focal length lens and is inclined.

[0008] The DMD fiber optic imaging module includes a digital micromirror device, a first plano-convex lens, a focusing objective lens, a beam expander objective lens, a second plano-convex lens, and a CCD. The digital micromirror device is located between the second positive focal lens and the first plano-convex lens;

[0009] The multimode square core fiber includes a fiber cladding and a square fiber core;

[0010] The control and acquisition module includes a data acquisition card and an industrial computer;

[0011] The deep learning image restoration algorithm module consists of a square mode speckle image, an image restoration neural network, and a restored grayscale image.

[0012] The wavelength of the fiber laser, the working band of the DMD fiber optic imaging module, and the detection wavelength of the CCD match, and the wavelength and working band are any one of the visible light band, the near-infrared band, or the short-wave infrared band.

[0013] The working bands of the fiber collimator, the reflector, the first positive focal lens, and the second positive focal lens match the wavelength of the incident fiber laser.

[0014] The working bands of the first plano-convex lens, the focusing objective lens, the beam expander objective lens, and the second plano-convex lens match the wavelength of the fiber laser.

[0015] The output position of the focused spatial laser generated by the focusing objective lens is within the square fiber core of the incident end face of the multimode square core fiber.

[0016] Two output ports of the data acquisition card need to confirm that a collection process is completed before proceeding to the next round, so that the grayscale image playback of the DMD fiber optic imaging module and the square mode speckle image acquisition of the CCD do not drop frames.

[0017] The image restoration neural network is a network with feature extraction and image restoration functions, and the network adopts any one of UNet, ResNet, and VGGNet.

[0018] The size of the spatial laser after beam expansion by the spatial light beam expander module is used to cover the size of the grayscale image played on the DMD fiber optic imaging module.

[0019] The technical effects achieved by the present invention are:

[0020] A multimode square core fiber imaging system of the present invention uses a multimode square core fiber as a core imaging element. Compared with traditional multimode fibers, the accuracy of restoring grayscale images is higher, and it can effectively solve the problem of circular contour background noise in traditional multimode fiber imaging, and has great application prospects in information transmission and endoscopic imaging based on multimode fibers. Description of the Drawings

[0021] Figure 1 is a schematic structural diagram of an embodiment of the present invention;

[0022] Figure 2 is a grayscale image restored by the multi-mode square core fiber imaging scheme of the embodiment of the present invention and a comparison diagram of the restoration results with a traditional multi-mode fiber.

[0023] Reference numerals:

[0024] 1. Spatial light beam expanding module; 101. Fiber laser; 102. Fiber collimator; 103. Reflecting mirror; 104. First positive focal lens; 105. Second positive focal lens; 2. DMD fiber imaging module; 201. Digital micromirror device; 202. First plano-convex lens; 203. Focusing objective lens; 204. Beam expanding objective lens; 205. Second plano-convex lens; 206. CCD; 3. Multi-mode square core fiber; 301. Fiber cladding; 302. Square fiber core; 4. Control and acquisition module; 401. Data acquisition card; 402. Industrial control computer; 5. Deep learning image restoration algorithm module; 501. Square mode speckle image; 502. Image restoration neural network; 503. Restored grayscale image. Detailed implementation manners

[0025] In order to make the objectives and advantages of the present invention clearer, the present invention will be specifically described below in conjunction with embodiments. It should be understood that the following text is only used to describe one or several specific implementation manners of the present invention, and does not strictly limit the specific protection scope claimed by the present invention.

[0026] As Figure 1 - Figure 2 shown, a multi-mode square core fiber imaging system includes a spatial light beam expanding module 1, a DMD fiber imaging module 2, a multi-mode square core fiber 3, a control and acquisition module 4, and a deep learning image restoration algorithm module 5;

[0027] The spatial light beam expanding module 1 includes a fiber laser 101, a fiber collimator 102, a reflecting mirror 103, a first positive focal lens 104, and a second positive focal lens 105. The reflecting mirror 103 is located between the fiber collimator 102 and the first positive focal lens 104 and is inclined;

[0028] The DMD fiber imaging module 2 includes a digital micromirror device 201, a first plano-convex lens 202, a focusing objective lens 203, a beam expanding objective lens 204, a second plano-convex lens 205, and a CCD 206. The digital micromirror device 201 is located between the second positive focal lens 105 and the first plano-convex lens 202;

[0029] The multi-mode square core fiber 3 includes a fiber cladding 301 and a square fiber core 302; the multi-mode square core fiber 3 is a multi-mode fiber with a square core;

[0030] The control and acquisition module 4 includes a data acquisition card 401 and an industrial personal computer 402;

[0031] The deep learning image restoration algorithm module 5 consists of a square pattern speckle image 501, an image restoration neural network 502, and a restored grayscale image 503. The deep learning image restoration algorithm module 5 is a schematic process diagram for using deep learning tools to perform grayscale image restoration based on the square speckles of the multimode square core fiber 3;

[0032] The wavelength of the fiber laser 101, the working band of the DMD fiber imaging module 2, and the detection wavelength of the CCD 206 match, and the wavelength and working band are any one of the visible light band, the near-infrared band, or the short-wave infrared band.

[0033] The working bands of the fiber collimator 102, the reflector 103, the first positive focal lens 104, and the second positive focal lens 105 match the wavelength of the incident fiber laser 101.

[0034] The working bands of the first plano-convex lens 202, the focusing objective lens 203, the beam expander objective lens 204, and the second plano-convex lens 205 match the wavelength of the fiber laser 101.

[0035] The output position of the focused spatial laser generated by the focusing objective lens 203 is within the square fiber core 302 of the incident end face of the multimode square core fiber 3.

[0036] The two output ports of the data acquisition card 401 need to complete one acquisition process before proceeding to the next round, so that the grayscale image playback of the DMD fiber imaging module 2 and the acquisition of the square pattern speckle image 501 by the CCD 206 do not drop frames.

[0037] The image restoration neural network 502 is a network with the function of feature extraction and image restoration, and the network adopts any one of UNet, ResNet, and VGGNet.

[0038] The size of the spatial laser after beam expansion by the spatial light beam expander module 1 is used to cover the size of the grayscale image played on the DMD fiber imaging module 2.

[0039] The laser emitted by the fiber laser 101 in the spatial light beam expansion module 1 is incident on the fiber collimator 102 through the optical fiber, and the laser transmitted by the optical fiber is collimated into the laser transmitted in space. The propagation path of the spatial laser is adjusted by the reflector 103, and the beam is expanded by the beam expander group composed of the first positive focus lens 104 and the second positive focus lens 105, and then it is injected into the DMD fiber imaging module 2; the spatial laser after beam expansion is irradiated on the digital micromirror device 201 at a specific incident angle, and the different grayscale images played on the digital micromirror device 201 are reflected. After the image intensity is modulated, it is incident on the first plano-convex lens 202, and the spatial laser is converged to a certain extent and then emitted to the focusing objective lens 203. The focusing objective lens 203 focuses the incident spatial laser and couples it into the multi-mode square core optical fiber 3. The incident spatial laser generates multi-mode interference in the multi-mode square core optical fiber 3, thereby generating a mode speckle output. The square mode speckle image 501 outputted from the square optical fiber core 302 in the output end face of the multi-mode square core optical fiber 3 is expanded by the beam expanding objective lens 204, and then emitted to the focusing objective lens 203 through the second plano-convex lens 20 5 is converged and shaped to a certain extent and then incident on CCD206 for detection. The industrial computer 402 interacts with CCD206 to collect the detected square pattern speckle image 501. In addition, the industrial computer 402 interacts and controls the data acquisition card 401 during the collection process. The two output ports of the data acquisition card 401 trigger the digital micromirror device 201 and CCD206 at the same time to ensure that each grayscale image played by the digital micromirror device 201 can be detected by CCD206. A corresponding square pattern speckle image 501 is thus formed. In the deep learning image restoration algorithm module 5, the image restoration neural network 502 is used to perform the grayscale image restoration task. In the training process, the collected data set is input into the image restoration neural network 502 to update the weight of the neural network. After the training is completed, the image restoration neural network 502 is used to restore the grayscale image of each input square pattern speckle image 501 to obtain the corresponding restored grayscale image 503.

[0040] The above is only a preferred embodiment of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention shall be implemented according to the conventional means in the art unless otherwise specified and limited.

Claims

1. A multimode square core optical fiber imaging system, characterized in that: It includes a spatial light beam expansion module (1), a DMD optical fiber imaging module (2), a multi-mode square core optical fiber (3), a control and acquisition module (4), and a deep learning image restoration algorithm module (5); The spatial light beam expansion module (1) comprises a fiber laser (101), a fiber collimator (102), a reflector (103), a first positive focus lens (104) and a second positive focus lens (105); the reflector (103) is located between the fiber collimator (102) and the first positive focus lens (104) and is distributed in an inclined manner; The DMD optical fiber imaging module (2) comprises a digital micromirror device (201), a first plano-convex lens (202), a focusing objective lens (203), a beam expanding objective lens (204), a second plano-convex lens (205) and a CCD (206), wherein the digital micromirror device (201) is located between the second positive focus lens (105) and the first plano-convex lens (202); The multimode square core optical fiber (3) comprises an optical fiber cladding (301) and a square optical fiber core (302); The control and acquisition module (4) comprises a data acquisition card (401) and an industrial computer (402); The deep learning image restoration algorithm module (5) is composed of a square pattern speckle image (501), an image restoration neural network (502) and a restored grayscale image (503).

2. A multimode square core optical fiber imaging system according to claim 1, characterized in that: The wavelength of the fiber laser (101), the operating band of the DMD fiber imaging module (2), and the detection wavelength of the CCD (206) match each other, and the wavelength and the operating band are any one of the visible light band, the near infrared band, or the short-wave infrared band.

3. A multimode square core optical fiber imaging system according to claim 1, characterized in that: The working wavelength bands of the optical fiber collimator (102), the reflector (103), the first positive focus lens (104) and the second positive focus lens (105) match the wavelength of the incident optical fiber laser (101).

4. A multimode square core optical fiber imaging system according to claim 1, characterized in that: The operating wavelength bands of the first plano-convex lens (202), the focusing objective lens (203), the beam expanding objective lens (204), and the second plano-convex lens (205) match the wavelength of the optical fiber laser (101).

5. The multimode square core optical fiber imaging system according to claim 1, characterized in that: The focusing spatial laser output position generated by the focusing objective lens (203) is within the square optical fiber core (302) at the incident end face of the multi-mode square core optical fiber (3).

6. The multimode square core optical fiber imaging system according to claim 1, characterized in that: The two output ports of the data acquisition card (401) need to confirm that one acquisition process is completed before proceeding to the next round, so that the grayscale image playback of the DMD optical fiber imaging module (2) and the acquisition of the square pattern speckle image (501) of the CCD (206) do not drop frames.

7. The multimode square core optical fiber imaging system according to claim 1, characterized in that: The image restoration neural network (502) is a network with a feature extraction and image restoration function, and the network adopts any one of UNet, ResNet and VGGNet.

8. The multimode square core optical fiber imaging system according to claim 1, characterized in that: The size of the spatial laser beam after beam expansion by the spatial light beam expansion module (1) is used to cover the size of the grayscale image played on the DMD optical fiber imaging module (2).

Images