A method for suppressing defocused image noise based on digital holographic microscopy imaging technology

Abstract

The present invention discloses a method for suppressing defocused image noise based on digital holographic microscopic imaging technology, comprising the steps of: subtracting background noise by subtracting adjacent holograms; preliminarily removing flyover noise by a three-dimensional deconvolution algorithm; detecting the edge of a target in a reconstructed image slice, and extracting a background portion that does not contain the target; propagating the background portion to a reconstructed image slice to be processed, and removing crosstalk between reconstructed image slices. The present invention effectively suppresses the defocused image noise in the reconstructed image slice, enables the digital holographic microscopic imaging technology to obtain three-dimensional imaging of a target through a hologram, and expands the application of digital holography in three-dimensional imaging.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional imaging based on digital holographic microscopic imaging technology, and in particular to a method for suppressing defocused image noise. Background Art

[0002] Digital holographic microscopy technology can record all the light wave field information of the target in a two-dimensional hologram, obtain the target's reproduced image slices layer by layer through the numerical reconstruction algorithm, and reconstruct the target's spatial distribution or three-dimensional image by obtaining the grayscale isosurfaces of these slices. However, digital holographic microscopy technology is only applicable to the spatial distribution of small simple particles or the three-dimensional imaging of transparent particles. For example, Latychevskaia et al. [1] The spatial distribution of simulated spherical particles with a diameter of 4 μm was obtained by a three-dimensional deconvolution algorithm. For opaque targets with complex morphology and large size (such as 10-100 μm plankton), it is difficult to obtain three-dimensional imaging of the target due to the severe out-of-focus image noise in the reconstructed image slices. Eom et al. [2] When a kidney tissue is imaged in three dimensions using digital holographic microscopy technology, only some spatially distributed points can be obtained, and a complete three-dimensional morphology cannot be formed.

[0003] The defocused images in the reconstructed image slices of digital holographic microscopy technology mainly come from the crosstalk between the reconstructed image slices, that is, a part of the target is clearly imaged in one of the reconstructed image slices, but it is a defocused blurred image when propagated to other reconstructed image slices through the numerical reconstruction algorithm, and vice versa. This crosstalk makes each reconstructed image slice contain not only a clear image of a part of the target, but also defocused images of other parts of the target, making the edges of the parts that should have been clearly imaged difficult to identify. Therefore, it is necessary to study a method to suppress the crosstalk between the reconstructed image slices, so that each part of the target can be clearly imaged in the reconstructed image slices, and its edges can be clearly identified, thereby realizing three-dimensional imaging of the target. Summary of the invention

[0004] The present invention provides a method for suppressing out-of-focus image noise based on digital holographic microscopy imaging technology, which can effectively remove the crosstalk between reproduced image slices after numerical reconstruction of a hologram, so that the outline of the target in the reproduced image slices is clear, thereby achieving a complete and accurate three-dimensional image of the target from a hologram.

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

[0006] A method for suppressing defocused image noise based on digital holographic microscopy imaging technology comprises the following steps:

[0007] (1) Obtain a hologram of the target and subtract the background noise from the adjacent hologram without the target;

[0008] A single-mode laser is used to increase the coherence length, and a 0.2 mm cuvette is used to limit the depth of target imaging within the coherence length range, ensuring that the entire target can be restored through numerical reconstruction of the hologram.

[0009] Since the essence of hologram is interference pattern, it contains a lot of coherent noise and high-frequency fringes. In order to eliminate these background noises, a hologram without the target taken at a nearby time is selected as the background for subtraction.

[0010] (2) numerically reconstructing the hologram after noise reduction in step (1) to obtain a series of original reconstructed image slices, so that the entire target can be imaged in these slices;

[0011] (3) A pinhole with a diameter of 1 μm is placed at 0 mm, 0.5 mm, 1 mm, and 1.5 mm away from the focal plane of the objective lens, and a hologram of the pinhole is captured. A reconstructed image slice with out-of-focus image noise removed is obtained through a three-dimensional deconvolution algorithm;

[0012] Since the 3D deconvolution algorithm needs to reproduce the 3D point spread function distribution of the image space, a pinhole with a diameter of 1 μm is placed at equal intervals near the focal plane of the objective lens, and the pinhole is used to replace an object point. Holograms are taken when the pinhole is open and closed, respectively, and the hologram of the pinhole is obtained according to the method of hologram subtraction in step (1), and the pinhole 3D imaging of the reproduced image space, that is, the 3D point spread function, is obtained by numerical reconstruction.

[0013] (4) detecting the edge of the target in the reconstructed image slice described in step (3), segmenting and removing the target therein, and obtaining a background reconstructed image slice without the target;

[0014] Experiments show that the crosstalk between reconstructed image slices mainly comes from the background in adjacent slices. Therefore, by detecting the edge of the target in the reconstructed image slices and extracting the background without the target, the crosstalk between slices is suppressed.

[0015] (5) Using the background reconstructed image slice described in step (4), the out-of-focus image noise in the reconstructed image slice described in step (3) is further removed by the reconstructed image slice crosstalk suppression method to obtain the final reconstructed image slice. During the implementation process, the reconstructed image slice is divided into two halves, front and back, according to the midline of the target. Since the processing of the front and back halves is similar, the front half is taken as an example, and the specific process includes:

[0016] 5-1. Starting from the reconstructed image slice at the target midline, take one reconstructed image slice every m reconstructed image slices forward, and take a total of 3 as standard reconstructed image slices, which are denoted as P0, P2 and P1 in sequence, where P0 is the reconstructed image slice at the target midline, P1 is the frontmost reconstructed image slice, and the reconstruction distances of the three are l0', l2' and l1', respectively, l0'>l2'>l1'. Take m reconstructed image slices forward in front of P1 as the reconstructed image slice to be processed, denoted as P Di ,i=1,2,…,m, their reconstruction distances are:

[0017] l′ Di =l′1-i×dl′(i=1,2,…,m) (1)

[0018] Where dl' is the reconstruction distance interval between two adjacent reconstructed image slices;

[0019] 5-2. Perform target edge detection on the standard reconstructed image slices P0, P2 and P1, and segment and remove the targets to obtain corresponding background reconstructed image slices P0', P2' and P1';

[0020] 5-3. Use the numerical reconstruction algorithm to forward propagate the background reconstruction image slices P0', P2' and P1' described in step (2) to the reconstruction image slice P to be processed. Di The propagation distances are l0'-l Di ', l2'-l Di ', l1'-l Di ', get P0", P2" and P1". According to the following formula, remove P0', P2' and P1' to P Di The crosstalk generated is:

[0021]

[0022] Where P Di (x,y), P0"(x,y), P2"(x,y) and P1"(x,y) are P Di , P0", P2" and P1". The longer the forward propagation distance, the more the defocused image noise diverges, and the smaller the crosstalk generated. For this reason, weight factors are set for P0"(x, y), P2"(x, y) and P1"(x, y), from small to large. Let i = i + 1, repeat steps 5-2 and 5-3, and remove the crosstalk between all m reconstructed image slices to be processed and the standard reconstructed image slices;

[0023] 5-4. Shift m reproduced image slices forward to obtain new standard reproduced image slices P0, P2 and P1 (at this time, P0 is the original P2, P2 is the original P1, and P1 is the original PD4 ), and reselect m to-be-processed reconstructed image slices. Repeat steps 5-2 to 5-4 to complete the suppression of crosstalk between the m to-be-processed reconstructed image slices and the standard reconstructed image slices in the next cycle, and the number of cycles is:

[0024]

[0025] Where l1' is the reconstruction distance of P1 in the first cycle, l min ' is the reconstruction distance of the first reproduced image slice, and '[]' is the rounding operator.

[0026] Beneficial effects of the present invention:

[0027] The present invention detects the edge of the target in the reconstructed image slice, extracts the background without the target, and uses the background to remove the crosstalk between the reconstructed image slices. And the defocus noise suppression effect is improved by processing in batches from the middle of the target to the head and tail ends;

[0028] The defocus image noise suppression method based on digital holographic microscopy imaging technology of the present invention effectively reduces the defocus image noise in the reproduced image slice, and obtains a clear outline of the entire target. The present invention can obtain three-dimensional imaging of the target through a hologram, so that digital holographic microscopy imaging technology has a broader application prospect in on-site monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Schematic diagram of the defocused image noise suppression method based on digital holographic microscopy imaging technology.

[0030] Figure 2 is a hologram containing a target.

[0031] Figure 3 is a hologram without a target.

[0032] Figure 4 This is the hologram with the background subtracted.

[0033] Figure 5 Slice the original reproduced image.

[0034] Figure 6 is the reconstructed image of the pinhole.

[0035] Figure 7 This is a reconstructed image slice with the out-of-focus noise initially removed.

[0036] Figure 8 To reproduce the edge of the object in the image slice.

[0037] Fig. 9 Reproduce image slices for background without target.

[0038] Fig.10 Schematic diagram of crosstalk suppression between reproduced image slices.

[0039] Fig.11 Slice the image for final reproduction.

[0040] Fig.12 This is a three-dimensional image of Heterobay akashiwo (HAZJ) reconstructed by connecting isosurfaces. DETAILED DESCRIPTION

[0041] The present invention will be described in detail below with reference to the embodiments and the accompanying drawings, but the present invention is not limited thereto.

[0042] In this embodiment, the hologram required for reconstructing and reproducing the image slice is obtained by shooting the coaxial optical path based on the Mach-Zehnder interferometer structure built in the laboratory. All algorithms are written in the environment of Matlab2018a, and the hardware conditions for the algorithm operation are core i5 processor, main frequency 2.6GHz, memory 4GB.

[0043] The schematic diagram of the method for suppressing defocused image noise based on digital holographic microscopy imaging technology in this embodiment is shown in FIG. Figure 1 As shown, the following steps are included:

[0044] (1) Continuously capture holograms of the target (in this case, Heterobay algae HAZJ) using a digital holographic microscopy imaging optical system (including a single-mode laser and a 0.2 mm thick cuvette), and select two holograms that contain the target and do not contain the target, respectively, at similar times, such as Figure 2 and 3 Subtract the two holograms to obtain the hologram with the background subtracted, as shown in Figure 4 As shown;

[0045] (2) 41 reconstruction distances are selected at equal intervals within the reconstruction distance range (11.1 mm ≤ l' ≤ 56.5 mm) to numerically reconstruct the hologram, and 41 original reconstructed image slices are obtained, such as Figure 5 As shown, the entire target can be imaged in these slices;

[0046] (3) In the direction away from the focal plane of the objective lens, a pinhole with a diameter of 1 μm is placed at 0 mm, 0.5 mm, 1 mm and 1.5 mm away from the focal plane, and the holograms of the pinholes are photographed in sequence and the corresponding reconstructed images of the pinholes are obtained by numerical reconstruction algorithms, such as Figure 6 As shown. According to the principle of proximity, the reconstructed image (i.e., point spread function) of the pinhole corresponding to each original reconstructed image slice is determined to obtain the three-dimensional point spread function distribution of the reconstructed image space. Finally, the reconstructed image slice with the out-of-focus image noise preliminarily removed is obtained through the three-dimensional deconvolution algorithm, as shown in Figure 7 As shown;

[0047] (4) Detecting the edge of the object in the reconstructed image slice described in step (3), such as Figure 8 The target is segmented and removed to obtain a background reconstruction image slice without the target, as shown in Fig. 9 As shown;

[0048] (5) Using the background reconstructed image slice described in step (4), the out-of-focus image noise in the reconstructed image slice described in step (3) is further removed by the reconstructed image slice crosstalk suppression method to obtain the final reconstructed image slice. During the implementation process, the reconstructed image slice is divided into two halves, front and back, according to the midline of the target. Taking the front half as an example, the specific process includes:

[0049] 5-1. Starting from the reconstructed image slice at the target midline, take one reconstructed image slice every m (in this example, m=4) reconstructed image slices, and take a total of three as standard reconstructed image slices, which are denoted as P0, P2 and P1 in sequence, where P0 is the reconstructed image slice at the target midline, and P1 is the frontmost reconstructed image slice. The reconstruction distances of the three are l0', l2' and l1' respectively (l0'>l2'>l1'), as shown in Fig.10 As shown. Take m reproduced image slices in front of P1 as reproduced image slices to be processed, denoted as P Di ,i=1,2,…,m, their reconstruction distances are:

[0050] l′ Di =l′1-i×dl′(i=1,2,…,m) (1)

[0051] Where dl' is the reconstruction distance interval between two adjacent reconstructed image slices;

[0052] 5-2. Perform target edge detection on the standard reconstructed image slices P0, P2 and P1, and segment and remove the targets to obtain corresponding background reconstructed image slices P0', P2' and P1';

[0053] 5-3. Use the numerical reconstruction algorithm to forward propagate the background reconstruction image slices P0', P2' and P1' described in step (2) to the reconstruction image slice P to be processed. Di The propagation distances are l0'-l Di ', l2'-l Di ', l1'-l Di ', get P0", P2" and P1". According to the following formula, remove P0', P2' and P1' to P Di The crosstalk generated is:

[0054]

[0055] Where P Di (x,y), P0"(x,y), P2"(x,y) and P1"(x,y) are P Di , P0", P2" and P1". The longer the forward propagation distance, the more the defocused image noise diverges, and the smaller the crosstalk generated. For this reason, weight factors are set for P0"(x, y), P2"(x, y) and P1"(x, y), from small to large. Let i = i + 1, repeat steps 5-2 and 5-3, and remove the crosstalk between all m reconstructed image slices to be processed and the standard reconstructed image slices;

[0056] 5-4. Shift m reproduced image slices forward to obtain new standard reproduced image slices P0, P2 and P1 (at this time, P0 is the original P2, P2 is the original P1, and P1 is the original P D4 ), and reselect m to-be-processed reconstructed image slices. Repeat steps (2) to (4) to complete the suppression of crosstalk between the m to-be-processed reconstructed image slices and the standard reconstructed image slices in the next cycle. The number of cycles is:

[0057]

[0058] Where l1' is the reconstruction distance of P1 in the first cycle, l min ' is the reconstruction distance of the first reproduced image slice, and '[]' is the rounding operator.

[0059] Fig.11 The final reconstructed image slice of HAZJ using the defocused image noise suppression method based on digital holographic microscopy imaging technology of the present invention is given. Figure 7 In contrast, the out-of-focus image noise is well suppressed by using the method of the present invention. The three-dimensional image of HAZJ reconstructed by isosurface connection is shown in FIG. Fig.12 As shown, a complete spatial profile is obtained.

[0060] The above description is only an example of a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for suppressing defocused image noise based on digital holographic microscopy imaging technology, characterized in that: Includes steps: (1) Obtain a hologram of the target and subtract the background noise from the adjacent hologram that does not contain the target; (2) numerically reconstructing the hologram after noise reduction in step (1) to obtain a series of original reconstructed image slices so that the entire target can be imaged in these slices; (3) Place a 1 mm diameter hole at 0 mm, 0.5 mm, 1 mm, and 1.5 mm away from the focal plane. μm A pinhole is used to capture a hologram of the pinhole, and a reconstructed image slice with out-of-focus image noise removed is obtained through a three-dimensional deconvolution algorithm; (4) detecting the edge of the target in the reconstructed image slice described in step (3), segmenting and removing the target therein, and obtaining a background reconstructed image slice without the target; (5) using the background reconstructed image slice described in step (4), further removing the out-of-focus image noise in the reconstructed image slice described in step (3) by a reconstructed image slice crosstalk suppression method, thereby obtaining a final reconstructed image slice; In step (5), the reconstructed image slice is divided into two halves, front and back, according to the midline of the target. For the front half, the specific steps of the reconstructed image slice crosstalk suppression method include: (5.1) Starting from the reconstructed image slice at the target midline, move forward every m Take one of the reconstructed image slices, and take a total of three as standard reconstructed image slices, which are denoted as P0, P2 and P1 respectively, where P0 is the reconstructed image slice at the target midline, P1 is the frontmost reconstructed image slice, and the reconstruction distances of the three are l 0', l 2' and l 1', l 0'> l 2'> l 1'; Continue forward in front of P1 m The reproduced image slices are taken as the reproduced image slices to be processed, denoted as P Di , i =1, 2, …, m , their reconstruction distances are: (1) where d l ' is the reconstruction distance interval between two adjacent reconstructed image slices; (5.2) Perform target edge detection on the standard reconstructed image slices P0, P2 and P1, and segment and remove the target to obtain the corresponding background reconstructed image slices P0', P2' and P1'; (5.3) Using the numerical reconstruction algorithm, forward propagate the background reconstruction image slices P0', P2' and P1' described in step (2) to the reconstruction image slice P to be processed. Di The propagation distances are l 0'- l Di '、 l 2'- l Di '、 l 1'- l Di ', get P0", P2" and P1"; remove P0', P2' and P1' according to the following formula Di The crosstalk generated is: (2) Where P Di (x,y), P0"(x,y), P2"(x,y) and P1"(x,y) are P Di , P0", P2" and P1" complex amplitude distribution. Since the longer the forward propagation distance, the more the defocused image noise diverges, the smaller the crosstalk generated. Therefore, weight factors are set for P0"(x,y), P2"(x,y) and P1"(x,y) from small to large. Let i=i +1, repeat steps (5.2) and (5.3) to remove all m Crosstalk between the reconstructed image slice to be processed and the standard reconstructed image slice; (5.4) Forward translation m The new standard reconstructed image slices P0, P2 and P1 are obtained (P0 is the original P2, P2 is the original P1, and P1 is the original P D4 ), and reselect m Repeat steps (5.2) to (5.4) to complete the next cycle. m The suppression of crosstalk between the reconstructed image slice to be processed and the standard reconstructed image slice, the number of cycles is: (3) in l 1' is the reconstruction distance of P1 in the first cycle, l min ' is the reconstruction distance of the first reproduced image slice, and '[]' is the rounding operator.

2. The method for suppressing defocused image noise based on digital holographic microscopy imaging technology according to claim 1, characterized in that: The step (1) is specifically as follows: (1) The target is photographed continuously through a digital holographic microscopy optical system to obtain a hologram of the target and a hologram without the target; the background noise is subtracted from the hologram of the target and the adjacent hologram without the target.

3. The method for suppressing defocused image noise based on digital holographic microscopy imaging technology according to claim 1, characterized in that: The step (2) is specifically as follows: (2) Select the reconstruction distance range according to the target size, and then numerically reconstruct the hologram after noise reduction in step (1).

Images