CN114265298A - Cell image recovery method for lens-free holographic imaging - Google Patents
Cell image recovery method for lens-free holographic imaging Download PDFInfo
- Publication number
- CN114265298A CN114265298A CN202111609687.5A CN202111609687A CN114265298A CN 114265298 A CN114265298 A CN 114265298A CN 202111609687 A CN202111609687 A CN 202111609687A CN 114265298 A CN114265298 A CN 114265298A
- Authority
- CN
- China
- Prior art keywords
- image
- propagation
- holographic imaging
- restored
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000003384 imaging method Methods 0.000 title claims abstract description 35
- 238000011084 recovery Methods 0.000 title description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 23
- 238000001228 spectrum Methods 0.000 claims description 3
- 238000011002 quantification Methods 0.000 claims description 2
- 230000000644 propagated effect Effects 0.000 claims 1
- 210000004027 cell Anatomy 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 230000001902 propagating effect Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 230000000750 progressive effect Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 2
- 241000195493 Cryptophyta Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000000601 blood cell Anatomy 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012123 point-of-care testing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
Images
Landscapes
- Holo Graphy (AREA)
Abstract
本发明公开了一种无透镜全息成像的细胞图像恢复方法。该方法的步骤如下：1：将图像传感器获取到的目标全息图反向传播到样本所在平面，得到称为反向传播图。2：对反向传播图执行阈值分割算法，得到阈值分割后的二值图像。3：对二值图像中的黑色部分执行最小外接圆算法，获得的最小外接圆作为初始支持域。4：固定初始支持域的圆心，以初始支持域的半径为基础向外扩展得到不同大小的同心圆，作为扩展支持域。5：分别以初始支持域和扩展支持域为基础进行图像恢复，分别得到不同支持域下的恢复图像。6：用方差量化各恢复图像的干涉条纹的波动性，取波动性最平缓的恢复图像，作为最优恢复图像。
The invention discloses a cell image restoration method of lensless holographic imaging. The steps of the method are as follows: 1: The target hologram acquired by the image sensor is back-propagated to the plane where the sample is located to obtain a back-propagation map. 2: Perform a threshold segmentation algorithm on the backpropagation graph to obtain a binary image after threshold segmentation. 3: Execute the minimum circumscribed circle algorithm on the black part of the binary image, and the obtained minimum circumscribed circle is used as the initial support field. 4: The center of the initial support field is fixed, and concentric circles of different sizes are obtained by expanding outward based on the radius of the initial support field, as the extended support field. 5: Perform image restoration based on the initial support domain and the extended support domain, respectively, and obtain restored images under different support domains. 6: Quantify the fluctuation of the interference fringes of each restored image by variance, and take the restored image with the smoothest fluctuation as the optimal restored image.
Description
Technical Field
The invention relates to the technical field of lens-free holographic imaging technology and holographic image recovery, in particular to a cell image recovery method for lens-free holographic imaging.
Background
Currently, the point-of-care (POCT) technology is rapidly developing. The instant diagnosis and treatment needs a portable, intelligent, rapid and miniaturized diagnosis and treatment detection device. Many instant diagnosis and treatment systems are developed on the basis of a microfluidic technology, a lens-free cell imaging technology and a deep learning technology, wherein the lens-free holographic imaging technology is widely applied, and the system has a large field of view, a high numerical aperture and richer recorded target (generally blood cell) information (phase information and amplitude information). Such systems typically record image information in the form of CMOS or CCD sensor values, followed by processing of the image using digital image processing techniques. For a holographic imaging system it records a holographic image of an object and in order to obtain the original appearance of the object it is generally necessary to process the hologram to reduce the original appearance of the object, a process generally referred to as image restoration.
In lensless holographic imaging systems, image recovery is a critical step in order to obtain a high resolution image of the target. A good image recovery algorithm needs to suppress noise and extract useful information as much as possible. The lens-free holographic imaging mode can introduce the phase loss problem and the twin image interference problem in the process of recording the holographic image by the image sensor. These two problems seriously affect the resolution of the final recovered image, so that the image recovery algorithm is generally required to have the capability of recovering the lost phase and suppressing the interference of the twin image, and the iterative phase retrieval algorithm is mainly adopted to solve the two problems at present. The algorithm is mainly based on an iterative recursive process by assuming the light field distribution of an image and propagating the light field among a recording plane (image sensor plane), a real image plane (object plane), and a virtual image plane (object plane symmetrically distributed with the sensor plane), and the iterative process is convergent and converges on the light field distribution of the object. Of course, the number of iterations is relatively large, so some constraints are introduced to accelerate convergence, generally, the amplitude of the light field is constrained by the amplitude recorded by the image sensor, and the distribution of the light field in the plane where the target is located is constrained by the area occupied by the target. The area occupied by this target is generally referred to as the support domain. The determination of the support domain is typically using a threshold segmentation algorithm.
However, the target of the current holographic imaging is generally semitransparent and low-contrast, such as a cell, and when the threshold segmentation algorithm is used for determining the support domain, the boundary of the target is often difficult to determine, so that a smaller support domain is segmented. In addition, the propagation distance from the sample to the image sensor, which is often on the order of microns, is used in the iterative propagation process for image recovery, and lensless holographic imaging systems are generally compact (do not use very sophisticated optics), meaning that an algorithm is required to accurately calculate this distance. Therefore, the two problems are solved, the image recovery quality of the lensless holographic microscope system can be effectively improved, and the image recovery speed can be increased.
Disclosure of Invention
The invention aims to optimize a method for dividing a support domain in an image recovery process of a lens-free holographic imaging system and a method for determining a propagation distance.
The invention provides a method for gradually exploring a support domain, which is used for solving the support domain information required by image restoration in a low-contrast target lensless holographic imaging process, and an automatic focusing algorithm which is used for determining the distance between a target and an image sensor in the lensless holographic imaging process (the distance parameter is one of key parameters in the image restoration step).
The steps of the method for progressively exploring the support domain are as follows:
step 1: and reversely propagating the target hologram acquired by the image sensor to the plane of the sample to obtain a reverse propagation diagram.
Step 2: and executing a threshold segmentation algorithm on the back propagation map to obtain a binary image after threshold segmentation.
And step 3: and executing a minimum circumcircle algorithm on a black part (with the gray value of 0) in the binary image, and taking the obtained minimum circumcircle as an initial support domain.
And 4, step 4: expanding a plurality of concentric circles with different radiuses of the initial support domain on the basis of the initial support domain to serve as an expanded support domain; the radius of each extended support domain is greater than the radius of the original support domain.
And 5: and respectively restoring the image on the basis of the initial support domain and the extended support domain to respectively obtain restored images in different support domains.
Step 6: the fluctuation of the interference fringes of each restored image is quantized by the variance, and the restored image with the most gradual fluctuation is taken as the optimal restored image. Compared with the prior art that the image obtained directly by threshold segmentation can reflect more details of the cells.
Further, the counter propagation in step 1 uses an angular spectrum propagation algorithm of plane waves for digital propagation.
Further, the threshold of the threshold segmentation algorithm in step 2 is obtained by using the Otsu method.
Further, after step 6 is executed, based on the radius corresponding to the obtained optimal recovery image, a new extension support field is added in a step smaller than the radius step in step 4; and restoring the image based on the new extended support domains to obtain a restored image. And 6, comparing the recovered image with the fluctuation of the interference fringes of the optimal recovered image determined in the step 6 to perform the optimal recovered image.
Further, the step length of radius expansion in step 4 is 1 pixel; the radius extension step size of the extended support field added after step 6 is performed is 0.1 pixel.
Further, the image restoration in step 5 is performed by using an iterative phase retrieval algorithm.
Further, the specific quantization mode for restoring the volatility of the interference fringes of the image in step 6 is as follows: and taking any point of the outer edge of the target in the restored image, drawing a perpendicular line of the outer edge of the restored image passing through the point, and calculating the variance of the gray distribution on the perpendicular line.
Further, in step 1, the distance of the target hologram in the backward propagation direction is obtained by an automatic focusing algorithm; the autofocus algorithm is specifically as follows:
and 1-1, taking the pure amplitude image as a target, and carrying out lens-free holographic imaging on the pure amplitude image to obtain a corresponding holographic image.
1-2, traversing the propagation distances in an estimation range by a certain step length, and reversely propagating the hologram by the propagation distances to obtain reverse propagation maps corresponding to different propagation distances.
1-3: and (3) executing a threshold segmentation algorithm on each back propagation image obtained in the step (1-2) to obtain a corresponding binary image.
1-4: calculating the number of white pixels (the gray value is 255) of the binary image obtained in the step 1-3; and taking the propagation distance corresponding to the binary image with the maximum number of white pixels as the optimal propagation distance value.
Further, the pure amplitude image described in step 1-1 was tested using a USAF standard resolution test plate.
Further, the traversal step length in the step 1-2 is 1 μm; after the optimal value of the propagation distance is obtained, fine traversal is performed on both sides of the optimal value of the propagation distance in steps of 0.2 μm, and the optimal value of the propagation distance is further updated.
Further, the traversal range in step 1-2 is 200 μm above and below the estimated value obtained by the hardware configuration of the original system.
Further, the threshold used by the threshold segmentation algorithm described in steps 1-3 is obtained by Otsu's method.
The invention has the beneficial effects that:
the optimization method provided by the invention is simple to operate and small in calculated amount, and can effectively improve the imaging resolution and robustness of the lens-free holographic imaging system (can better perform image recovery on a target with low contrast). In the imaging process of the lens-free holographic imaging system, the distance between an optimized target and an image sensor is determined by an automatic focusing algorithm, an optimized support domain is obtained by using a progressive exploration support domain method, and finally the image quality of an output image of the whole system is improved.
Drawings
FIG. 1 is a basic block diagram of lensless holographic imaging.
Fig. 2 is a flow chart of an autofocus algorithm of the present invention.
FIG. 3 is a flowchart illustrating the overall steps of the present invention.
Fig. 4 is a flowchart of acquiring an extended support domain in the present invention.
Fig. 5 is a diagram of the actual operation process of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the drawings in the specification.
As shown in fig. 1, a lensless holographic imaging system consists essentially of a light source, a pinhole, a sample plane, and a detection plane. The autofocus algorithm of the present invention is primarily used to determine the micrometer-scale distance between the sample plane and the detection plane, while the progressive support domain exploration method is primarily used to optimize the determination of the support domain when the hologram counter-propagates to the sample plane during the system image recovery process. The use of these two optimization methods in lensless holographic imaging is detailed below.
Firstly, before the lens-free holographic system images the target (generally cells, algae, micromechanical structures, etc.), a USAF standard resolution test plate is placed on the system sample plane (fig. 1), and the hologram of the resolution test plate is recorded. An autofocus algorithm is performed with the hologram as the object.
The flow of the auto-focusing algorithm is shown in FIG. 2, which first obtains the distance (Z) between the sample plane and the detection plane according to the hardware parameters of the system2) Estimate of (in general)<1mm) and traverse up and down to 200 μm above and below this estimate in steps of 1 μm, and then back-propagate each traverse to obtain a back-propagated image. Then, the Otsu method is combined with a threshold segmentation algorithm to obtain a segmentation result (binary image) of each back propagation map, and then the sum of pixel points of which the gray value is 255 of the binary image is calculated. The maximum point of the last sum corresponds to the optimum Z2The value is obtained. Z which can be obtained under 1 μm if higher accuracy is required2A finer traversal is performed within the neighborhood of (a).
Z derived using an autofocus algorithm2And obtaining a back propagation diagram under the condition, and carrying out the subsequent progressive exploration support domain method by using the back propagation diagram.
As shown in fig. 3, the method first performs a threshold segmentation algorithm combined with the ohs method on the target direction propagation map to obtain a basic support domain. And then, expanding the support domain outwards to obtain a series of support domains, and then obtaining recovery images corresponding to the conditions of the expanded support domains. And finally, evaluating the quality of the restored image by measuring the gray distribution variance of a perpendicular line from any point on the edge of the target image of the restored image to the outer edge of the restored image (the smaller the variance is, the better the image quality is), wherein the support domain corresponding to the restored image with the best image quality is optimal.
Fig. 4 shows specific contents of the extended support domain in the progressive exploration support domain method. Firstly, the minimum circumcircle of the basic support domain (called as initial support domain) is found, and then the fixed center of the circle is expanded by a certain step size to obtain a series of concentric circles (called as expansion support domain). This step size is typically 1 μm, although smaller steps may be used if a higher accuracy of the support field is desired.
The following provides a specific implementation of the present invention:
step 1: and reversely propagating the target hologram acquired by the image sensor to the plane of the sample to obtain a reverse propagation diagram. The back propagation uses the angular spectrum propagation algorithm of plane waves for digital propagation.
The distance of the target hologram in reverse propagation is obtained by an automatic focusing algorithm; the autofocus algorithm is specifically as follows:
and 1-1, taking the pure amplitude image as a target, and carrying out lens-free holographic imaging on the pure amplitude image to obtain a corresponding holographic image. The pure amplitude images were tested using a USAF standard resolution test plate.
1-2, traversing the propagation distances in an estimation range by a certain step length, and reversely propagating the hologram by the propagation distances to obtain reverse propagation maps corresponding to different propagation distances. The traversal step length is 1 μm; after the optimal value of the propagation distance is obtained, fine traversal is performed on both sides of the optimal value of the propagation distance in steps of 0.2 μm, and the optimal value of the propagation distance is further updated. The traversal range is 200 μm interval above and below the estimated value obtained by hardware configuration of the original system.
1-3: and (3) executing a threshold segmentation algorithm on each back propagation image obtained in the step (1-2) to obtain a corresponding binary image. The threshold used by the threshold segmentation algorithm is obtained by using the Otsu method.
1-4: calculating the number of white pixels (the gray value is 255) of the binary image obtained in the step 1-3; and taking the propagation distance corresponding to the binary image with the maximum number of white pixels as the optimal propagation distance value.
Step 2: and executing a threshold segmentation algorithm on the back propagation map to obtain a binary image after threshold segmentation. The threshold of the threshold segmentation algorithm is obtained by using the Otsu method.
And step 3: and executing a minimum circumcircle algorithm on a black part (with the gray value of 0) in the binary image, and taking the obtained minimum circumcircle as an initial support domain.
And 4, step 4: and fixing the center of the initial support domain, and outwards expanding the initial support domain to obtain concentric circles with different sizes as the expanded support domain on the basis of the radius of the initial support domain. The radius extension step is 1 pixel.
And 5: and respectively restoring the image on the basis of the initial support domain and the extended support domain to respectively obtain restored images in different support domains. And the image recovery is carried out by adopting an iterative phase retrieval algorithm.
Step 6: the fluctuation of the interference fringes of each restored image is quantized by the variance, and the restored image with the most gradual fluctuation is taken as the optimal restored image. The details of the interference fringes are shown in the figure. Compared with the prior art that the image obtained directly by threshold segmentation can reflect more details of the cells. The specific quantification method for recovering the volatility of the interference fringes of the image is as follows: and taking any point of the outer edge of the target in the restored image, drawing a perpendicular line of the outer edge of the restored image passing through the point, and calculating the variance of the gray distribution on the perpendicular line.
Step 7, on the basis of the corresponding radius of the obtained optimal recovery image, adding a new expansion support domain according to the radius step length smaller than that in the step 4; and restoring the image based on the new extended support domains to obtain a restored image. And 6, comparing the recovered image with the fluctuation of the interference fringes of the optimal recovered image determined in the step 6 to perform the optimal recovered image. The radius extension step size of the extension support field in this step is 0.1 pixel.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111609687.5A CN114265298B (en) | 2021-12-24 | 2021-12-24 | Cell image restoration method for lens-free holographic imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111609687.5A CN114265298B (en) | 2021-12-24 | 2021-12-24 | Cell image restoration method for lens-free holographic imaging |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114265298A true CN114265298A (en) | 2022-04-01 |
| CN114265298B CN114265298B (en) | 2024-02-09 |
Family
ID=80830192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111609687.5A Active CN114265298B (en) | 2021-12-24 | 2021-12-24 | Cell image restoration method for lens-free holographic imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114265298B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120218379A1 (en) * | 2009-10-20 | 2012-08-30 | The Regents Of The University Of California | Incoherent lensfree cell holography and microscopy on a chip |
| CN108508588A (en) * | 2018-04-23 | 2018-09-07 | 南京大学 | A kind of multiple constraint information without lens holographic microphotography phase recovery method and its device |
| CN112327473A (en) * | 2021-01-07 | 2021-02-05 | 南京理工大学智能计算成像研究院有限公司 | Lensless microscopic imaging system and image reconstruction method based on average projection iteration |
| CN113359403A (en) * | 2021-05-21 | 2021-09-07 | 大连海事大学 | Automatic focusing method for lens-free digital holographic imaging |
-
2021
- 2021-12-24 CN CN202111609687.5A patent/CN114265298B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120218379A1 (en) * | 2009-10-20 | 2012-08-30 | The Regents Of The University Of California | Incoherent lensfree cell holography and microscopy on a chip |
| CN108508588A (en) * | 2018-04-23 | 2018-09-07 | 南京大学 | A kind of multiple constraint information without lens holographic microphotography phase recovery method and its device |
| CN112327473A (en) * | 2021-01-07 | 2021-02-05 | 南京理工大学智能计算成像研究院有限公司 | Lensless microscopic imaging system and image reconstruction method based on average projection iteration |
| CN113359403A (en) * | 2021-05-21 | 2021-09-07 | 大连海事大学 | Automatic focusing method for lens-free digital holographic imaging |
Non-Patent Citations (3)
| Title |
|---|
| HU, JQ; LI, Q ； ZHOU, Y: "Support-domain constrained phase retrieval algorithms in terahertz in-line digital holography reconstruction of a nonisolated amplitude object", APPLIED OPTICS, vol. 55, no. 2, pages 379 - 386 * |
| 肖卓磊, 基于优化理论的相位恢复算法研究, pages 72 - 81 * |
| 蒋帅帅, 基于细胞图像的计算全息算法研究, pages 2 - 34 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114265298B (en) | 2024-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110781924B (en) | Side-scan sonar image feature extraction method based on full convolution neural network | |
| WO2022121531A1 (en) | Product defect detection method and apparatus | |
| CN112734915A (en) | Multi-view stereoscopic vision three-dimensional scene reconstruction method based on deep learning | |
| CN113221925A (en) | Target detection method and device based on multi-scale image | |
| WO2024066035A1 (en) | Defect detection method and system based on battery surface image, and related device | |
| Ali et al. | Robust focus volume regularization in shape from focus | |
| CN116485779B (en) | Adaptive wafer defect detection method, device, electronic equipment and storage medium | |
| CN114897738A (en) | Image blind restoration method based on semantic inconsistency detection | |
| Shen et al. | Deeper super-resolution generative adversarial network with gradient penalty for sonar image enhancement | |
| CN113158487A (en) | Wavefront phase difference detection method based on long-short term memory depth network | |
| Yan et al. | Multiscale fusion and aggregation pcnn for 3d shape recovery | |
| CN115359108A (en) | Depth prediction method and system based on defocusing under guidance of focal stack reconstruction | |
| CN114998358A (en) | Multi-focus image fusion method and device, computer equipment and storage medium | |
| CN114265298A (en) | Cell image recovery method for lens-free holographic imaging | |
| Fernandes et al. | Improving focus measurements using logarithmic image processing | |
| WO2024227398A1 (en) | Image processing method for gene sample, image processing apparatus, and storage medium | |
| CN118623769A (en) | A method for detecting the width of a pole piece material of a photoelectric sensor | |
| Mutahira et al. | Jitter noise modeling and its removal using recursive least squares in shape from focus systems | |
| CN112200831A (en) | Dense connection twin neural network target tracking method based on dynamic template | |
| CN115358962B (en) | End-to-end visual odometer method and device | |
| CN117557720A (en) | Microscopic three-dimensional morphology reconstruction method of lightweight depth network | |
| CN114529476B (en) | Phase retrieval method for lensless holographic microscopy based on decoupling-fusion network | |
| CN114372944B (en) | Multi-mode and multi-scale fused candidate region generation method and related device | |
| Rajagopalan et al. | A recursive algorithm for maximum likelihood-based identification of blur from multiple observations | |
| CN115393641A (en) | Immunofluorescence image classification method and device, equipment, storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |