312 results about "Phase image" patented technology

The invention discloses a method and a device for correcting laser focusing aberration in a transparent material, and belongs to the technical field of laser processing. The method comprises the steps that a machining light path containing an object plane, an aspheric reflector and a 4F device is built in Zemax, the surface type of the aspheric reflector is optimized with the minimum focus aberration as the target, and a rise table is derived; an optical path difference and a laser phase are calculated through Matlab, and a phase diagram is generated; and an actual machining light path is built, and a phase diagram is loaded to achieve low-aberration machining. Simulation optimization and phase modulation are combined, aberration caused by refractive index difference is effectively counteracted, wavefront errors are reduced by 58%, the thickness of a machining damage layer is reduced by 70%, the method is suitable for various transparent materials, the machining precision and the material utilization rate are improved, and the method is suitable for high-precision laser machining scenes.

The invention provides a water chlorophyll a concentration inversion method and system based on multi-modal data and a lightweight model, and relates to the technical field of water environment remote sensing evaluation. The method comprises the following steps: firstly, acquiring a Gaofeng No.5 satellite remote sensing image, a sentinel No.3 satellite image and ground actual measurement data, and completing image preprocessing and water body pixel extraction; constructing a hyperspectral index and an aquatic vegetation index, and fusing the hyperspectral index and the aquatic vegetation index with the water body temperature, the pH environmental factors and the spectral reflectivity to form a multi-dimensional feature sample set; a core feature subset is obtained through random forest and XGBoost coupling feature selection, and a lightweight student model is trained based on knowledge distillation; and constructing a to-be-predicted feature sample for the to-be-predicted time phase image and the environment factor, inputting the to-be-predicted feature sample into the lightweight student model to obtain a chlorophyll a concentration predicted value, and generating a spatial distribution map and a quality control map layer. According to the invention, high-precision, low-redundancy and efficient deployment chlorophyll a concentration inversion is realized.

The invention provides a random forest shallow sea sediment classification method based on multi-temporal remote sensing image fusion, and relates to the technical field of sediment information extraction. Comprising the following steps: 1, collecting and preprocessing multi-temporal image data to obtain remote sensing reflectivity; 2, the water depth of each single-time-phase image is inverted, and the optimal water depth is obtained; 3, calculating bottom reflectivity characteristics of blue and green wave bands based on the optimal remote sensing image; 4, respectively calculating topographic features and spectral features based on the optimal water depth and the optimal remote sensing image; and 5, in combination with the bottom reflectivity features, the topographic features and the spectral features, carrying out random forest feature optimization and classification model training, and generating a substrate classification result. On the basis, the method solves the problems that an existing remote sensing image substrate classification method is insufficient in feature consideration, noise in a single-time-phase image can cause low classification precision, and therefore negative effects can be generated on accurate acquisition of substrate information.

The invention provides a structured light projector calibration method and system and a medium, and the method comprises the steps: projecting a multi-frequency stripe pattern to a target region based on an MEMS micromirror array module, obtaining a distorted image sequence, and carrying out the phase decoding, and obtaining a phase diagram corresponding to each frequency; resolving the phase diagram based on a phase shift method and an optical path triangulation principle to obtain a preliminary three-dimensional point coordinate; constructing a neural network model and a calibration sample set, obtaining a real three-dimensional coordinate by using a standard calibration plate or a known curved surface, and calculating the deviation between the initial three-dimensional point coordinate and the real three-dimensional coordinate to obtain a training sample set; and training the neural network model according to a training sample set based on an Euclidean distance loss function, outputting a three-dimensional offset compensation amount according to the neural network model, correcting the initial three-dimensional point coordinates, and realizing an end-to-end error self-compensation mechanism through the neural network model, so that deviation compensation is effectively performed on the three-dimensional point coordinates, and the accuracy of the three-dimensional point coordinates is improved. And the three-dimensional reconstruction precision is improved.

The invention relates to the field of image processors and biological medicines, and discloses an NK cell activity rapid detection method based on image processing. The method comprises the following steps: acquiring an unmarked time sequence phase image sequence of an NK cell and target cell co-culture system; performing cell instance segmentation to track individual cells; extracting a morphological dynamic characteristic parameter set of the target cell, wherein the morphological dynamic characteristic parameter set comprises a volume change rate, a phase gradient entropy, a cytoplasm phase fluctuation frequency and a nuclear region phase mean value; inputting the parameters into a pre-trained death state discrimination model, and outputting a death probability; and calculating a killing efficiency index based on the death probability evolution curve, and judging the activity level of the NK cells. The system comprises a phase image acquisition unit, a cell segmentation unit, a feature extraction unit, a death judgment unit and an activity judgment unit. Through unmarked imaging and deep learning fusion analysis, high-precision, real-time, quantitative and ultra-early NK cell activity evaluation is realized, and the method is suitable for clinical instant inspection and immunotherapy monitoring.

The invention provides a Flash wafer detection method and device based on deep learning, and relates to the technical field of semiconductor detection. The method comprises the following steps: firstly, receiving an initial electric signal response sequence which is acquired by a scanning probe and comprises reflection intensity and phase deviation; performing time-frequency domain joint transformation processing on the sequence to generate a frequency domain energy distribution map and a time domain attenuation characteristic curve; further constructing a three-dimensional space mapping model containing a frequency domain energy amplitude and a time domain attenuation time constant; a pre-trained condition generation network is called to reconstruct the model, and a reconstructed phase image set with the resolution consistent with that of a standard template is generated; and finally, performing pixel-by-pixel comparison on the reconstructed image and a standard template to generate a detection report containing defect space positioning coordinates and a contour boundary sequence. According to the method, deep mining and visual reconstruction of Flash wafer microscopic electric signal features are realized, and the accuracy and the automation degree of defect detection are remarkably improved.

The invention discloses a laser shearing speckle interference phase unwrapping method based on deep learning. According to the method, eight speckle patterns before and after deformation are used as input, multi-scale feature extraction is carried out through layer-by-layer convolution and down-sampling of an encoder, features are sent to an attention fusion module, response weights of channels and spatial positions are adaptively adjusted through channel attention and position attention, then the response weights are input into a decoder, and jump connection with an encoding end is combined, so that multi-scale feature fusion is realized. And step-by-step reconstruction of multi-scale features is realized, and a phase diagram is output. Performing comprehensive constraint on prediction and reference phases in the aspects of numerical deviation, structural consistency and gradient smoothness and updating network parameters by adopting composite loss formed by mean square error, mean absolute error, structural similarity, gradient loss and out-of-plane displacement calculation items; therefore, an unwrapping result which is globally continuous and has clear and stable local details is obtained, and a feasible way is provided for intelligent processing and automatic analysis of the laser shearing speckle interference image.

The invention discloses a building remote sensing image change detection method based on a twin network and an attention mechanism, and the method comprises the steps: extracting a feature sequence of a building remote sensing image at the same place at different times through employing the twin network, and adding the attention mechanism of a spatial dimension to obtain a multi-scale fusion feature. The method comprises the following steps: firstly, preprocessing a dual-time-phase remote sensing image and constructing a data set; then, an improved twin neural network model is constructed, the model embeds attention mechanisms in an encoder and a decoder, and multi-scale features of the dual-time-phase image are fused; training the model by adopting a mixed loss function; and finally, carrying out change detection by utilizing the trained model and outputting a result. According to the method, feature extraction consistency is guaranteed through the twin network, attention to a building change area is enhanced by using an attention mechanism, large-scale and detail features are highlighted by using feature fusion, the precision and robustness of change detection are effectively improved, and the method is suitable for building change monitoring in a complex environment.

The invention relates to the technical field of remote sensing change detection, and discloses a remote sensing change detection method and system based on boundary perception semantic context. The method comprises the following steps: extracting global spatio-temporal features of a double-temporal image pair under different scales through a boundary perception semantic context network model to obtain time-phase differential features of each scale, extracting texture and contour information from the time-phase differential features of each scale to obtain detail features, and integrating the time-phase differential features of each scale to obtain semantic features; and introducing the boundary information into a learning process, fusing the detail features and the semantic features to obtain enhanced features, and performing prediction according to the boundary information and the enhanced features to obtain a remote sensing change detection result. According to the invention, information of different scales and boundary information in the remote sensing image can be effectively utilized, and the accuracy of remote sensing change detection is improved.

The invention discloses a microfluidic image flow cytometry identification system and a microfluidic image flow cytometry identification method based on multi-modal phase imaging and deep learning, relates to a microfluidic image flow cytometry detection technology, and belongs to the crossing field of microfluidics, optical imaging and artificial intelligence. The cell deformation chip is used for inducing a cell sample to generate controllable deformation through the cell deformation chip and discharging deformed cells; the optical imaging unit is used for scanning the deformed cells and collecting multi-modal phase images of the deformed cells; the data processing and analysis unit is used for preprocessing the multi-modal phase image and performing cell intelligent identification analysis on the preprocessed multi-modal phase image by using a deep learning feature extraction network, so that accurate identification and classification of cells are realized, and a leukocyte subpopulation identification model is constructed; and generating a cell imaging result and a cell mechanical parameter thermodynamic diagram. According to the invention, label-free identification of leukocyte subgroups is realized, and the problems of tedious operation and cell damage of traditional fluorescence labeling are solved.

The invention relates to the technical field of cell culture, and discloses a stem cell differentiation degree identification method based on image features. The method comprises the following steps: continuously capturing multi-temporal image data in a stem cell culture process through a high-resolution microscopic imaging system, and generating an image and environment synchronous data set; processing the data set, extracting dynamic morphological features and texture change modes of the stem cells, and constructing a fusion feature set; outputting a differentiation process index based on the fusion feature set; a differentiation degree grade division rule is set, and a differentiation process index and differentiation grade mapping table is established; in a real-time application stage, acquiring real-time imaging data and environment readings, and inputting the real-time imaging data and the environment readings into an evaluation network to calculate real-time differentiation process indexes; querying the mapping table according to the real-time index to obtain a differentiation grade judgment result; comparing the judgment result with a preset target range, and if the judgment result falls into the target range, automatically executing differentiation state marking and culture parameter adjusting operation.

The invention provides a split-phase cooperative asphalt mixture digital image multi-component segmentation method, and relates to the field of asphalt mixture image segmentation, and the method comprises the steps: obtaining a digital image of an asphalt mixture section, converting the digital image into a gray image, and carrying out the preprocessing of the gray image, and obtaining a to-be-segmented image; performing gap segmentation on the to-be-segmented image to obtain a gap phase image mask; segmenting the to-be-segmented image by using the RAN-UNet deep learning image segmentation model to obtain an aggregate image mask; performing three-dimensional voxel reconstruction to obtain a three-dimensional digital model, and performing connectivity analysis and edge detection to obtain a connected gap mask, an open gap mask and a closed gap mask; performing critical particle size screening and structural gap identification to obtain a structural gap mask and a mortar internal gap; and carrying out particle size screening on the aggregate image mask, extracting an asphalt mortar region, and slicing and converting the asphalt mortar region again to obtain a mortar phase image. According to the invention, the segmentation precision of the asphalt mixture digital image is improved.

The application provides a complex terrain area elevation inversion method based on small unmanned aerial vehicle borne interferometric SAR. N-track small unmanned aerial vehicle flight tracks are set, the X-axis and Y-axis coordinates of the starting point and ending point of each track small unmanned aerial vehicle are unchanged, and the value of the Z-axis is gradually increased along the vertical direction. N-track echo data are obtained by the small unmanned aerial vehicle borne SAR according to the set tracks, and BP algorithm is used for imaging processing to obtain N SAR images. The offset of the SAR images of adjacent short baseline small unmanned aerial vehicles is obtained. The offset of the SAR images of long baseline small unmanned aerial vehicles is calculated based on the offset of the SAR images of small unmanned aerial vehicles, and the SAR images of long baseline are image matched. The SAR images after image matching are processed to obtain corresponding interference phase images, and the interference phase images are filtered by using Goldstein algorithm. Multi-baseline phase unwrapping is performed based on the interference phase of the interference phase images. The complex terrain area elevation inversion method of SAR is realized.

The present application relates to the technical field of industrial surface defect detection, and particularly relates to a phase deflection detection method based on asymmetric sampling and multi-scale gradient fusion, which comprises the following steps: firstly, extracting several phase shift phase images in horizontal and vertical directions, and using a region interpolation algorithm to down-sample and reduce dimensions to reduce calculation amount and suppress noise; based on a four-step phase shift method, demodulating wrapped phase and calculating multi-class reflection component images, and completing phase unwrapping through vectorization one-dimensional path accumulation; for different image types, using bicubic, nearest neighbor and bilinear differential up-sampling schemes to restore original resolution; performing small, medium and large three-scale gradient extraction on unwrapped phase, introducing weight coefficients to fuse features through a weighted maximum absolute value method; finally, generating two types of defect detection maps of gradient amplitude and gradient direction; the present application can greatly reduce calculation cost, and simultaneously consider different scale defect detection.

The application discloses a kind of diffractive light neural network computing systems and methods for realizing pure optical nonlinearity, belong to photonic computing and artificial intelligence hardware technical field.Its system includes coherent light source, linear light computing unit, nonlinear light activation unit and optical detection unit arranged in order along optical path, wherein linear light computing unit is programmable phase modulation using spatial light modulator, nonlinear light activation unit is programmable binary amplitude modulation using digital micromirror device, and both are directly coupled to form pure optical computing path without photoelectric conversion on optical path.Its method trains neural network through customized loss function, drives nonlinear activation function output to binary convergence, and maps the parameters obtained by training into phase map and binary mask respectively and loads to hardware.The application realizes true all-optical nonlinear computation, with the advantages of high energy efficiency, extremely low delay, compact structure and strong parallel processing capability.

The invention discloses a neurology interventional operation risk assessment method based on artificial intelligence, and belongs to the technical field of medical information, and the method specifically comprises the steps: obtaining a cerebrovascular multi-phase image and vascular wall biomechanical data of a target patient; reconstructing a three-dimensional blood vessel model based on the image, extracting an area change curve of each segment of lumen, and constructing a blood loss harmonic network according to transmission delay and morphological difference between the curves; distributing fluid-solid coupling simulation parameters containing an individualized vascular wall compliance curve for each vascular segment based on biomechanical data; planning an instrument path, and simulating interaction of an instrument and a blood vessel wall to generate a local pressure pulse sequence; inputting the sequence into a blood flow resonance loss network, and calculating blood flow resonance intensity and phase shift cumulant caused at a specific network node; and recognizing a high-risk blood vessel region with the delayed blood flow redistribution disorder according to the high-risk blood vessel region, and outputting a structured list. According to the invention, quantitative evaluation and early warning of networked dynamic risks caused by interventional operations are realized.

The present application relates to a kind of remote sensing image change detection method and equipment based on multiscale difference feature fusion, belong to remote sensing image processing technical field.The present application is by designing multi-attention feature enhancement module to double time-phase image and its difference information are adaptively weighted and enhanced, while using cross-scale residual fusion module to the feature from different scales is unified fusion, global change feature after fusion is input into classification head and is carried out change detection, the accurate capture of the detail and overall structure of the change region of ground object is realized, to effectively improve the accuracy and robustness of change detection, and also can obtain better detection performance under the condition of less training sample.

The application discloses an optical image encryption and decryption method based on deep learning and random mask coding. In the encryption process, the encrypted image is passed through an optical device with two parallel optoelectronic couplers, and the encoding of the encrypted image is realized through random masks at different positions, so that two speckles as ciphertexts are obtained at the receiving end of the optical device. Before decryption, the two collected speckles are respectively sparsified by using two different sparse matrices, and then combined together, and the two sparse matrices are used as part of the decryption key. In the decryption process, the trained neural network model is used as another part of the key, and then the combined speckles and the pure phase image are respectively used as the input and label of the trained neural network model, and the decrypted image is output by the trained neural network model. The application can effectively encrypt images and has good robustness and security.

This invention discloses a three-dimensional contour measurement method and apparatus based on multi-projection gratings, relating to the field of three-dimensional topography measurement technology. The method includes: obtaining intrinsic and extrinsic parameters through calibration; projecting a set of speckle patterns; preliminarily analyzing the surface topography gradient of the object to determine a non-uniform spatial frequency sequence; sequentially projecting and acquiring multiple sets of phase-shifting grating patterns, and calculating the wrapping phase map corresponding to each frequency; for the same pixel, based on its reliability assessment in wrapping phase maps at different frequencies, assigning confidence weights to the phase values ​​of each frequency, and calculating and optimizing the phase map and the corresponding fused confidence map; calculating gradient information and combining it with the fused confidence map to perform region division, and executing a region-guided path consistency verification unfolding algorithm to obtain the absolute phase; converting the absolute phase into three-dimensional point cloud coordinates of the object surface. This application aims to solve the problems of complex topography, discontinuities, and noise easily leading to unfolding path errors and order jumps during phase unfolding.

The application discloses a phase solving method based on sequence pixel extreme value positioning, relates to the fields of computer vision and optical measurement, and comprises the following steps: designing M line shift or phase shift fringe patterns and N complementary gray code patterns, projecting the patterns to an object to be measured in sequence, collecting images of the object to be measured under different fringe patterns, and obtaining M line shift fringe images or phase shift fringe images and N complementary gray code images; performing phase calculation on the collected M line shift fringe images or phase shift fringe images, and obtaining a wrapped phase image; decoding the collected first N-1 gray codes to obtain K1 orders, and decoding all the N gray codes to obtain K2 orders; combining the wrapped phase image and the multiple gray code projection images to perform phase unwrapping, and obtaining a phase unwrapping image; and through the innovative extreme value positioning and adaptive exposure processing mechanism, the accuracy, the robustness and the reliability of the phase solving are effectively improved, and the problem that the measurement effect of the prior art is poor in a complex environment is solved.

The invention relates to the technical field of optical fiber sensing and optical fiber device manufacturing, and discloses a femtosecond laser-inscribing polarization-maintaining fiber grating method and system for vector sensing, and the method comprises the following steps: obtaining a stress region image of a polarization-maintaining fiber through a CCD camera, recognizing the slow axis direction of the fiber through image processing, rotating the fiber, and obtaining the stress region image of the polarization-maintaining fiber; an included angle of 45 degrees is formed between the slow axis direction and a preset incident plane of the femtosecond laser; calculating a conjugate phase of wavefront aberration generated when femtosecond laser penetrates through a fiber coating layer and a cladding and is focused to a fiber core, and generating a phase diagram for loading a spatial light modulator; after being modulated by the spatial light modulator, a light beam emitted by the femtosecond laser enters and is focused on the fiber core along the direction of an included angle of 45 degrees, and a grating structure is formed by combining axial scanning and inscribing. Accurate automatic alignment in the fast and slow axis direction and active compensation of transmission aberration are achieved, refractive index modulation birefringence forming a specific included angle with the inherent birefringence axis of the optical fiber can be effectively induced, and therefore the polarization-maintaining fiber grating with the highly controllable reflection spectrum and the asymmetric double-peak characteristic is engraved.

The present application relates to the field of information optics, more particularly, to a spatial structure vertical focal field generation method and system. The present application provides a spatial structure vertical focal field generation method, according to the constructed phase map of the spatial structure vertical focal field, the incident light beam is modulated into a cylindrical wave light beam at the light beam modulation, and then the cylindrical wave light beam is focused to generate a spatial structure vertical focal field. Wherein, the phase map only needs to be calculated according to the superposition of the column lens phase or the column lens phase and the flash grid phase, and the calculation amount and calculation time are obviously reduced compared with the iterative calculation of the existing method. The present application solves the problems of large calculation amount and long time consumption of the existing method for generating a spatial structure vertical focal field.

The invention relates to a method and a device for identifying multi-class dispersed phases in a multiphase flow system, which relate to the technical field of multiphase flow dispersed phase image processing, and adopt a plurality of independent and special deep learning detection models to carry out parallel detection on the same frame of image; based on the physical priori knowledge of each target, asymmetric filtering rules (different confidence thresholds, sizes and length-width ratio constraints) customized for each model are applied to the initial output of each model; meanwhile, a conflict arbitration mechanism based on the space overlapping degree and the preset category priority is introduced, and the overlapping targets recognized in a cross-model mode are precisely judged so as to eliminate recognition ambiguity, that is, the technical route of'special model detection, asymmetric screening and conflict arbitration 'is adopted, and therefore the recognition accuracy of the overlapping targets is improved. And the accuracy, the reliability and the result uniqueness of online detection of multiple types of dispersed phases in a complex multiphase flow system are remarkably improved.

The invention relates to the technical field of ultra-low field magnetic resonance systems, in particular to a battery charge state detection method and system based on closed-loop adaptive ultra-low field MRI, and the method comprises the steps: collecting the phase change of a water molecule H signal caused by the magnetic field disturbance of a to-be-detected battery, and reconstructing a first signal phase image; extracting to obtain a high-information-content area mask, and then performing resampling to obtain a second signal phase image; and inputting the second signal phase image into a state-of-charge detection model to predict a state-of-charge value of the to-be-detected battery. In order to solve the problem that a battery magnetic resonance detection scheme in the prior art needs to adopt a high-field magnetic resonance system and is not beneficial to production line use, detection equipment is replaced with an ultralow-field magnetic resonance system, so that structures such as a shielding room in a traditional magnetic resonance system are omitted, and a high-information-content area is positioned firstly, so that the detection efficiency is improved; and resampling is carried out on the region, so that the imaging precision is improved, and the accuracy of state of charge estimation of the detection model is improved.

The application relates to a kind of quantitative imaging methods of living cells based on digital holographic microscopic sequential displacement aberration correction, and belongs to the field of optical measurement, comprising: collecting three holograms by sequentially displacing living cells; obtaining three original phase maps by phase demodulation; obtaining total mask map; sequentially displacing phase map in reverse direction; performing difference calculation to obtain global phase system aberration difference data; obtaining local system aberration difference data by using total mask map; performing difference processing on Chebyshev polynomial to obtain local polynomial difference data by using total mask map; fitting global system aberration by calculating corresponding coefficients from difference data; subtracting global system aberration from three original phase maps to obtain phase map of living cells after aberration correction. The application realizes high-precision quantitative phase imaging of living cells while retaining high-frequency information, has excellent displacement distance robustness, does not need to be equipped with high-precision displacement mechanism, and greatly reduces the hardware cost of measurement system.

Methods and systems for image feature classification use NN-based learning algorithms to make decisions about features in medical images or image portions. In particular, embodiments can leverage phases of a multi-phase image to improve classification accuracy. For example, embodiments can combine different phases of a multi-phase image as training data.

A composite material gas cylinder layering defect displacement prediction method based on deep learning and laser shearing speckle interference comprises the following steps: constructing a training data set combining simulation data and experimental data, generating the simulation data based on a shearing speckle interference imaging principle, and combining the experimental data acquired by an experiment to predict the layering defect displacement of a composite material gas cylinder. Constructing a mixed data set containing a wrapped phase diagram and an out-of-plane displacement diagram; constructing a deep convolutional neural network fused with an attention mechanism, and extracting displacement features from high-frequency stripes: starting from an input layer, sequentially connecting an encoder path, a bottleneck layer, a decoder path and an output layer; constructing a composite loss function and training the network, and constructing a composite loss function comprising a pixel-level error constraint and a gradient consistency constraint; and utilizing the trained deep learning neural network to predict out-of-plane displacement response at the layering defect of the gas cylinder. According to the method, the problems of image distortion, data missing and the like caused by a traditional phase unwrapping algorithm are effectively solved, and the out-of-plane displacement field can be accurately, smoothly and continuously reconstructed.

The invention discloses a geometric phase metasurface high-flux laser direct writing system and method, and the system comprises a femtosecond laser, a beam expander, a angular dispersion compensation module, a digital micromirror array, a second 4F system, a spatial light modulator, a fifth lens and an objective lens according to the light advancing direction. Wherein the spatial light modulator is used for performing phase modulation on each unit light spot from the digital micromirror array based on a loaded holographic phase diagram to form a rod-shaped light spot array, and the rod-shaped light spot array forms parallel light transmitted at different angles through the fifth lens and converges at an entrance pupil of the objective lens together. And finally, a parallel direct-writing light spot array is formed in the two-photon photoresist, and geometric phase metasurface unit structure array printing is completed. According to the system, high-flux and high-flexibility direct-writing manufacturing of the geometric phase type metasurface is finally realized.

The invention discloses a region change detection method and system based on unmanned aerial vehicle vision, and the method comprises the steps: constructing a region change detection model through employing images of the same region in different periods, and enabling the region change detection model to capture a dynamic evolution mode (such as gradual vegetation degradation or sudden building extension) of the surface change; the characteristic limitation of traditional single-time phase or double-time phase comparison is broken through; a double-time-phase image pair is generated through a real-time image based on unmanned aerial vehicle vision, it is ensured that the double-time-phase image pair is analyzed under a unified coordinate system and a spectrum reference, the detection stability is improved, and therefore the anti-jamming capability is improved; through the full-process closed-loop design from data division, model training to real-time detection, in combination with the end-to-end reasoning capability of a deep learning model, the manual intervention demand is reduced, manual threshold setting or feature screening is not needed, change pattern spots are directly output in the test stage, and full automation from data input to result output is achieved.

The invention discloses a digital differential interference imaging method, and belongs to the technical field of optical imaging and microscopic imaging. The method comprises the steps that bright field illumination is used, a first image and a second image are collected at defocus positions which are vertically symmetrical on a sample focal plane respectively, and the defocus distance is larger than the physical depth of field of an objective lens; and carrying out pixel-level gray value subtraction on the two images to obtain a digital differential phase image. Digital image processing is adopted to replace optical shearing and interference of a traditional DIC, a polarized optical element and a special DIC prism are not needed, the system cost and complexity are remarkably reduced, the system can be compatible with non-standard optical material sample containers such as a plastic culture dish, and the directional shadow problem of the traditional DIC is solved. Meanwhile, by providing two specific implementation modes, namely accurate positioning acquisition and automatic focusing fusion acquisition, the imaging efficiency and the system robustness are further improved on the basis of ensuring the high contrast and the pseudo three-dimensional relief imaging effect.