Method for carrying out single shooting imaging on target in thick scattering medium by using near-infrared light

Abstract

The invention discloses a method for single shooting and imaging of a target in a scattering medium based on near-infrared light. According to the method, the statistical property of the speckles and the conclusion that the root cutting number of the covariance of the speckles is equal to the autocorrelation amplitude of the target are utilized to realize single shooting imaging of the target in the thick scattering medium based on the near-infrared light, and meanwhile, the imaging of the target in the fresh-cut chicken breast tissue of 20 average free paths can be realized at the deepest; and the maximum imaging resolution is 150 [mu] m. According to the method, the image can be reconstructed clearly only by shooting the speckle of the corresponding target, so that the calculation amount in the aspect of scattering imaging is greatly reduced; near-infrared light is used as a light source, so that the imaging depth is improved, the maximum average free path can reach 20, and a new thought is provided for application in the fields of biomedicine and the like.

Description

Technical Field

[0001] The present application relates to the technical field of scattering imaging, and in particular to a method for high-resolution single-shot infrared imaging through a thick scattering medium based on speckle correlation. Background Art

[0002] In a scattering medium, the incident plane light wave will cause the light wavefront to be distorted due to the inhomogeneity of the medium's refractive index. Especially when the imaging target is located inside the scattering medium, the difficulty and complexity of imaging are greatly increased. How to effectively image targets hidden in scattering media has always been an important challenge in the field of optical imaging. To address this problem, researchers have proposed a variety of imaging methods. Early studies used ultrafast time-gate technology to locate hidden targets and verified the feasibility of imaging in scattering media. Subsequently, cameras based on single-photon time-of-flight can quickly image targets in strong scattering media without long-term sampling. Using polarized light detection technology, some studies have also achieved the separation of ballistic photons from background scattered light, thereby successfully imaging targets in turbid media. However, many methods are limited by thick scattering media and are difficult to be widely used. In order to overcome this limitation, some studies have used technologies such as fluorescent labeling, compressed sensing, and spatial light modulation to improve imaging resolution and depth, but these methods still require taking a large number of speckle images and performing complex data processing. In addition, the Fourier shower curtain effect and deep learning-based techniques have also been proposed to achieve deep imaging in scattering media, but most of them still rely on a large number of speckle image acquisitions. In recent years, new methods such as unlabeled semi-supervised learning, space-time encoding (STEP), and photoacoustic imaging have also been proposed, further promoting the development of scattering medium imaging technology. These methods have improved imaging depth, speed, and resolution to varying degrees, but still face some challenges, such as high sampling requirements and adaptability to thick scattering media.

[0003] This study proposed an imaging method based on near-infrared light source and speckle statistics, aiming to solve the problem of single-frame imaging of targets hidden in thick scattering media, especially to achieve deeper penetration and imaging in thicker biological tissues or scattering media. Summary of the invention

[0004] In order to achieve high-resolution rapid imaging of targets inside thick scattering media, the present invention proposes a method for single-shot imaging of targets inside thick scattering media based on near-infrared light. Through the reasoning of the theoretical model, it is known that the square root of the covariance of the speckle is equal to the amplitude of the autocorrelation of the target. Then we use the phase retrieval algorithm (HIO) to restore the image and obtain a reconstructed image with high fidelity. The present invention builds an experimental optical path with near-infrared light as the light source, using a 1550-nanometer near-infrared fiber laser as the light source through an isolator and a fiber amplifier, and then connecting to a fiber collimator, and then irradiating the near-infrared light to the scattering medium with the target embedded inside. After that, the camera captures the speckle of the target on the exit surface of the scattering medium. We read the captured speckle with Matlab and then perform a phase retrieval algorithm for covariance calculation to obtain a reconstructed image. The present invention finally achieves the reconstruction of the target embedded in the chicken breast tissue of 20 mean free paths, and can achieve a maximum resolution of 150μm.

[0005] A method for single-shot imaging of targets inside scattering media based on near-infrared light; The theoretical model is the core of single-shot imaging. The theoretical model is inspired by the book "Speckle Phenomenon in Optics". When a plane light wave is irradiated on a scattering medium, the light passes through scattering medium 1 and irradiates the target in the form of speckle illumination. The light intensity distribution on the target surface can be decomposed into multiple point light sources. The photons emitted by each point light source pass through scattering medium 2 in a random path. Finally, the photons meet again on the back surface of scattering medium 2. Photons with an optical path difference less than the coherence length interfere with each other to form speckles. This process can be regarded as an impulse response. Then the speckle pattern is received by a short-wave infrared camera. According to the description of the statistical characteristics of speckle in the book, we deduce that the square root of the covariance of the speckle is equal to the amplitude obtained by the target autocorrelation, so that the covariance of the speckle can be input into the phase retrieval algorithm to restore and reconstruct the target embedded in the scattering medium through iteration. Therefore, we can use the speckle of a target to achieve image reconstruction based on this theoretical model.

[0006] A system for single-shot imaging of targets inside scattering media based on near-infrared light; The system for single-shot imaging of targets inside scattering media based on near-infrared light includes a 1550nm fiber laser as a light source, which is amplified by a fiber amplifier through a fiber isolator and then connected to a fiber laser collimator for emission. The 1550nm near-infrared is irradiated onto the scattering medium after collimation, and then emitted from the rear surface of the scattering medium, and the speckle pattern of the target is captured by a short-wave infrared camera.

[0007] The beneficial effects of the present invention are: The present invention uses the conclusion that the square root of the covariance of the speckle is equal to the amplitude of the autocorrelation of the target to realize single-shot imaging of the target inside the thick scattering medium based on near-infrared light, and can realize the imaging of the target inside the fresh-cut chicken breast tissue with a depth of 20 mean free paths at most, and the maximum imaging resolution is 150μm. The present invention only needs to shoot one speckle of the corresponding target to reconstruct the image clearly, which greatly reduces the amount of calculation in scattering imaging and provides new ideas for its application in biomedicine and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is the schematic diagram of the theoretical model.

[0009] Figure 2 The process of reconstructing an image using a single speckle. a. The speckle pattern of the target captured by the camera; b. The covariance diagram of the speckle; c. The amplitude information of the target is obtained by taking the square root of the covariance of the speckle and then fed into the HIO algorithm to obtain the reconstructed image of the target; d. The original image of the reconstructed image captured by the camera.

[0010] Figure 3 Experimental optical path diagram for imaging targets inside scattering media.

[0011] Figure 4 For reconstruction of different targets embedded in thin scattering media.

[0012] Figure 5 Zinc oxide is used as a scattering medium to image the target inside it.

[0013] Figure 6 Imaging the inside of a thick scattering medium (chicken breast tissue). a. The overall thickness of the chicken breast is 3mm, with 2mm in front of the target and 1mm behind the target; b. The overall thickness of the chicken breast is 4mm, with 3mm in front of the target and 1mm behind the target; c. The overall thickness of the chicken breast is 5mm, with 4mm in front of the target and 1mm behind the target; Bar: 1mm.

[0014] Figure 7 Imaging of the inside of chicken breast tissue. a. The thickness of the chicken breast is 4mm, 3mm before the target, and 1mm after the target; b. The overall thickness of the chicken breast is 5mm, 3mm before the target, and 2mm after the target; c. The overall thickness of the chicken breast is 6mm, 3mm before the target, and 3mm after the target; Bar: 1mm. DETAILED DESCRIPTION

[0015] The present invention is further described below by way of embodiments in conjunction with the accompanying drawings.

[0016] Figure 1.a shows the physical theoretical model of this method. First, when a plane light wave is irradiated on Diffuser1, it passes through Diffuser1 and then irradiates the target input surface O in : Formula 1 here is the field at the output surface of Diffuser1, and then the light is output by the target output surface O out Output: Formula 2 The speckle field generated by Diffuser1 is equivalent to speckle illumination, so the amplitude A and phase in the above formula are It is caused by the speckle field. After the light passes through the second scattering medium, the second exit surface of the scattering medium is set as: Formula 3 The light is then received by the CCD camera. Formula 4 The uniform speckle pattern captured by the camera is as follows: Figure 1 .b; Therefore, we can write the speckle captured by the camera as follows: Formula 5 Then we perform a covariance calculation on the obtained speckle: Formula 6 The autocorrelation of speckle can be written as: Formula 7 Here C represents the normalized covariance of the speckle, and * represents the autocorrelation operation. According to formulas (6) and (7), we can conclude that the square root of the covariance of the speckle is equal to the amplitude of the autocorrelation of the target O. After obtaining the amplitude information of the target, we send it to the phase retrieval algorithm (HIO). After 250 iterations, the reconstructed target image is obtained. The image reconstruction process is as follows: Figure 2 .

[0017] like Figure 3 As shown, Figure 3The experimental optical path diagram of the proposed method is shown. The target used in the experiment is letters and numbers engraved on a metal plate by laser drilling. The 1550nm near-infrared light emitted by a fiber laser (Optilab, Benchtop DFB Source) passes through a fiber isolator (Thorlabs, Fiber Isolator, 1550nm, Low Power, FC / APC) and enters a fiber amplifier (Connet, Mars Fiber Amplifier). The optical power is then amplified to 50mW and connected to a fiber collimator (Thorlabs, Singlemode GRIN Fiber Collimator, 1550nm, FC / APC). The collimated beam is then irradiated onto the scattering medium embedded in the target. The generated speckle is captured by a short-wave infrared camera (TEKWIN). SYSTEMS, SC640); the scattering medium embedded in the target in the experiment was made by clamping the target tightly together with two pieces of scattering medium, and the scattering surfaces of scattering medium one and scattering medium two were both close to the target; the scattering media used were 120-grain (Thorlabs, DGUV10-120) and 220-grain round glass diffusers (DGUV-220), 450μm thick zinc oxide (zinc oxide was plated on a 2×2cm thick quartz glass sheet), and chicken breast tissue (1mm, 2mm, 3mm, and 4mm respectively); the short-wave near-infrared camera had a pixel of 512×640, a single pixel size of 15μm, and the distance between the camera and scattering medium two was 18cm.

[0018] A total of four sets of experiments were carried out based on the proposed method to verify the effectiveness and efficiency of the proposed method.

[0019] Usually we use the mean free path to determine the thickness of the scattering medium. The mean free path is equal to the inverse of the scattering coefficient. When the thickness of a scattering medium is greater than the mean free path, we consider the scattering medium to be a thick scattering medium.

[0020] The first set of experiments is to use the proposed method to image different targets embedded in thin scattering media; the second, third and fourth sets of experiments are to use the proposed method to image different targets embedded in different thick scattering media at different distances.

[0021] like Figure 4 As shown, this figure shows the experimental results of the first set of experiments.

[0022] The first set of experiments used a set of 120-grain and 220-grain round glass diffusers. The target was sandwiched between the scattering surfaces of the two diffusers and was close to each other. From left to right in the figure, there are four columns, namely the speckle pattern of the target captured by the camera, the covariance of the target speckle, the reconstructed image, and the original image of the target captured by the camera. A total of four targets of different shapes were used in the experiment, namely 2, Y, E, and 37, with a target size of 2×3mm. All targets were recovered from the speckle of a single target. From the figure, we can clearly distinguish the shape and details of the recovered target. Therefore, from this set of experimental data, we analyzed and verified that our theoretical model can use a single speckle to recover the target embedded in the scattering medium.

[0023] like Figure 5 As shown, this figure shows the second set of experiments, in which zinc oxide is used as a scattering medium to image the target inside it.

[0024] The second set of experiments replaced the scattering medium with 450μm zinc oxide (zinc oxide coated on a quartz glass sheet of 2×25×25mm size) and fresh-cut chicken breast tissue of different thicknesses. First, we used two pieces of 450μm zinc oxide as the scattering medium (so that the overall mean free path of the two scattering media reached about ten), and placed the side coated with zinc oxide close to the input and output surfaces of the target. The camera was 17cm away from the second scattering medium to capture the target's speckle. The four columns in the figure are the target's speckle, the covariance reconstructed target, and the original image taken by the camera; all images are restored from a speckle of the corresponding target. Comparing the reconstructed image with the experiment of the thin scattering medium, it was found that the clarity of the restored image was basically the same.

[0025] like Figure 6 , Figure 7 As shown, the images show the results of the third and fourth experiments on imaging through chicken breast tissue.

[0026] In the experiment, the target was embedded in the chicken breast tissue, and the thickness of the chicken breast was 3mm, 4mm, 5mm, and 6mm respectively. First, we embedded the target 1mm away from the rear surface of the scattering medium and obtained Figure 6 A set of results. The figure shows a total of three sets of results. The first column of each set of results is the original image of the target taken by the camera, and the second column is the image restored using a single speckle of the corresponding target. The thickness of the scattering medium in these three sets of experiments is gradually increased in steps of 1mm. From the experimental results, it can be analyzed that for single-digit or letter targets, relatively clear images can be restored, but for multi-target '37', except for Figure 6(a) The restored image can be clearly distinguished. The restoration results of the remaining two groups of experiments gradually decrease. Overall, when the distance from the front surface of the target to the front surface of the scattering medium reaches 4 mm, the image restoration quality of the optical system has reached its limit. At this time, the entire scattering medium has about 15 mean free paths.

[0027] In the fourth group of experiments, we still used chicken breast as the scattering medium, kept the thickness from the front surface of the target to the front surface of the scattering medium at 3mm, and gradually increased the thickness from the back surface of the target to the back surface of the scattering medium in steps of 1mm. The experimental results are shown in Figure 2. Figure 7 As shown in the figure, there are three sets of experimental results. The first column of each set of results is the original image of the target captured by the camera, and the second column is the restored image. From the experimental results, we can see that for targets with single letters or numbers, a clearer image can be restored, but for targets with two numbers, only an outline and some details can be restored.

Claims

1. A method for single-shot imaging of a target inside a thick scattering medium based on near-infrared light, characterized in that: The invention comprises a theoretical model for realizing single shot image reconstruction and a single shot imaging optical path, wherein the theoretical model is used to explain that a single shot can be used to realize restoration and reconstruction of an image of a target inside a thick scattering medium, and the imaging optical path is used to realize single shot imaging of a target inside a scattering medium.

2. The method for single-shot imaging of a target inside a thick scattering medium based on near-infrared light according to claim 1, characterized in that: The theoretical model includes the application of covariance theory based on speckle statistical characteristics in the theoretical model, and it is deduced through speckle statistical characteristics and covariance theory that the square root of the covariance of the speckle is equal to the amplitude of the target autocorrelation, and then the image is reconstructed through a phase retrieval algorithm.

3. The method for single-shot imaging of a target inside a thick scattering medium based on near-infrared light according to claim 2, characterized in that: In the image reconstruction process, after the covariance of the speckle is obtained by calculation, it is sent to the phase retrieval algorithm (HIO) and finally the reconstructed image is obtained after 250 iterations. The entire image restoration process is implemented on MATLAB.

4. The method for single-shot imaging of a target inside a thick scattering medium based on near-infrared light according to claim 1, characterized in that: The imaging optical path includes an optical fiber isolator, an optical amplifier, an optical fiber collimator, a scattering medium embedded in a target, and an infrared short-wave camera.

5. The method for single-shot imaging of a target inside a thick scattering medium based on near-infrared light according to claim 3, characterized in that: The scattering medium embedded in the target is a combination of 120-grain and 220-grain scattering sheets whose scattering surfaces are close to the front and rear surfaces of the target, and a combination of two quartz glass sheets coated with 450μm zinc oxide tightly clamping the target. The target is embedded in the scattering medium of 3mm, 4mm, 5mm, and 6mm chicken breast tissue respectively.

Images