Quantum dot short-wave infrared detector non-uniformity calibration and correction method and system

Abstract

The application belongs to the technical field of photoelectricity, and specifically discloses a quantum dot short-wave infrared detector non-uniformity calibration and correction method and system. The calibration method comprises the following steps: determining a plurality of calibration conditions set in advance; in each calibration condition, acquiring different original input response values of each pixel unit to different preset light intensity light signal inputs; and determining ideal input response values of all pixel units to each preset light intensity light signal input; in each calibration condition, fitting a calibration function between the original input response values of each pixel unit to different light signals and the corresponding ideal input response values, and determining a calibration model corresponding to each pixel unit; the calibration model is used for correcting the original input response values of the pixel unit, obtaining a corrected input response, so that the corrected input response matches the ideal input response value. Through the application, the non-uniformity correction accuracy of the detector image is improved.

Description

Technical Field

[0001] This application belongs to the field of optoelectronic technology, and more specifically, relates to a method and system for non-uniformity calibration and correction of a quantum dot shortwave infrared detector. Background Technology

[0002] Compared to traditional InGaAs short-wave infrared detectors, quantum dot short-wave infrared detectors typically achieve short-wave infrared response by integrating quantum dot photosensitive films with readout circuits. These detectors offer advantages such as low cost, large-area fabrication capability, integration with silicon-based readout circuits, and spectral response controllable through quantum dot materials and size, making them suitable for low-cost, high-resolution short-wave infrared imaging applications. Publicly available research also indicates that colloidal quantum dot short-wave infrared detectors feature adjustable bandgap through quantum dot size and monolithic integration with silicon readout integrated circuits (ROICs) via solution processing.

[0003] However, the imaging consistency of quantum dot short-wave infrared detectors is also affected by factors such as the uniformity of quantum dot film deposition, material energy level matching, carrier transport, dark current, interface defects, inter-pixel response differences, and temperature drift. Especially in low-illuminance, long integration times, high-gain, or temperature-varying scenarios, quantum dot short-wave infrared images are more prone to exhibiting spatial variation noise, fixed pattern noise, dark field shadows, response nonlinearity, and condition-related inhomogeneities. Existing short-wave infrared imaging studies have also indicated that short-wave infrared (SWIR) images typically exhibit significant spatial variation noise, including dark shadows or fixed pattern noise.

[0004] Existing shortwave infrared modules on the market typically use a two-point correction method to handle image non-uniformity. This involves calculating the gain and bias of each pixel through dark and bright field calibration, and then performing a linear mapping at runtime. The traditional two-point linear correction model is as follows: in, Indicates pixel index, Indicates the integration time as Time Pixel The original input response, Indicates the correction output. and Representing pixels The gain coefficient and bias coefficient.

[0005] This method is simple to implement and has low hardware overhead, but it assumes that each pixel approximately satisfies a linear response within its working range. Its adaptability is poor because the non-uniformity of quantum dot short-wave infrared images varies significantly with different exposure times, focal plane temperatures, and gain modes. A single set of correction data is insufficient to adapt to complex dynamic environments, leading to noticeable image variations in different outdoor scenes and impacting user experience. When the integration time varies significantly, the pixel response curve exhibits significant non-linearity, or the dynamic range of different pixels differs greatly, a single gain bias matrix cannot cover all operating conditions, easily resulting in residual fixed pattern noise, local brightness mismatch, and image style drift. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this application is to provide a non-uniformity calibration and correction method and system for quantum dot shortwave infrared detectors. This aims to solve the problem of poor non-uniformity correction adaptability of quantum dot shortwave infrared detectors under different integration times, gain modes, and focal plane temperatures, caused by differences in the response of quantum dot photosensitive films, dark current and spatial variation noise, nonlinear response between pixels, and condition-related drift.

[0007] To achieve the above objectives, in a first aspect, this application provides a method for non-uniformity calibration of a quantum dot short-wave infrared detector, wherein the quantum dot short-wave infrared detector comprises multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity of the detected original image. The method includes: Multiple pre-set calibration conditions are determined; the detector's conditions include at least one of integration time, gain, and focal plane temperature. Under each calibration condition, acquire the different original input response values ​​of each pixel unit to light signal input with different preset light intensities; and determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. Under each calibration condition, a calibration function is fitted between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals to determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response so that the corrected input response matches the ideal input response value.

[0008] This application adopts a pixel-by-pixel polynomial modeling method, which can more accurately describe the nonlinear response relationship of each pixel of the quantum dot shortwave infrared detector under different brightness and integration time conditions, thereby improving the non-uniformity correction accuracy of the quantum dot shortwave infrared detector.

[0009] In one possible implementation, the method further includes: Determine the current operating status of the detector; From the plurality of calibration conditions, determine at least two calibration conditions that are adjacent to the current calibration condition's parameters; Based on the interval of operating parameters between the current operating condition and each adjacent calibration operating condition, the calibration model of the corresponding pixel unit under each adjacent calibration operating condition is interpolated to obtain the calibration model of each pixel unit under the current operating condition.

[0010] In one possible implementation, when the detector's operating conditions include only one of integration time, gain, and focal plane temperature, linear interpolation is performed on the calibration models of corresponding pixel units under two adjacent calibration conditions to obtain the calibration models of each pixel unit under the current operating condition. And / or when the detector operating condition includes at least two of integration time, gain, and focal plane temperature, multidimensional interpolation is performed on the calibration models of corresponding pixel units under each adjacent calibration condition based on the relative positions of the operating parameters of the current operating condition on the operating axis corresponding to different operating conditions, so as to obtain the calibration model of each pixel unit under the current operating condition.

[0011] In one possible implementation, the calibration function is a polynomial function; and / or the basis functions of the polynomial model are Legendre basis functions.

[0012] In one possible implementation, fitting the calibration function includes: The original input response value of each pixel unit to different light signals is normalized to obtain the normalized original input response value; The ideal input response value of each pixel unit to different light signals is normalized to obtain the normalized ideal input response value; Using the normalized original input response value as the input value and the normalized ideal input response value as the output value, the polynomial function is fitted to determine the coefficients of each order of polynomial, thus obtaining the corresponding calibration model.

[0013] In one possible implementation, determining the ideal input response value for all pixel units to each preset light intensity light signal input includes: Acquire T detection images of the detector for the same preset light intensity light signal input, where T is a positive integer greater than 1; Determine the mean input response for each detected image; The ideal input response value of the preset light intensity signal is obtained by averaging the T input responses corresponding to the T detection images.

[0014] Secondly, this application provides a method for correcting the non-uniformity of a quantum dot short-wave infrared detector. The quantum dot short-wave infrared detector includes multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity in the original detected image. The method includes: Determine the current operating condition of the detector and obtain the raw input response of each pixel unit under the current operating condition; The calibration method described in the first aspect or any possible implementation of the first aspect is used to obtain the calibration model of each pixel unit under the current working condition; Based on the calibration model of each pixel unit, the original input response of each pixel unit is non-uniformity corrected to obtain the corrected output image.

[0015] Thirdly, this application provides a non-uniformity calibration system for a quantum dot short-wave infrared detector. The quantum dot short-wave infrared detector includes multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity of the original detected image. The system includes: The calibration condition determination module is used to determine multiple pre-set calibration conditions; the detector's conditions include at least one of integration time, gain, and focal plane temperature. The input response determination module is used to acquire the different original input response values ​​of each pixel unit to light signal input with different preset light intensities under each calibration condition; and to determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. The calibration model determination module is used to fit a calibration function between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals under each calibration condition, and determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response, so that the corrected input response matches the ideal input response value.

[0016] Fourthly, this application provides a non-uniformity correction system for a quantum dot short-wave infrared detector. The quantum dot short-wave infrared detector includes multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity in the detected original image. The system includes: The raw response determination module is used to determine the current operating condition of the detector and to obtain the raw input response of each pixel unit under the current operating condition. The calibration model determination module is used to obtain the calibration model of each pixel unit under the current working condition by adopting the calibration method described in the first aspect or any possible implementation of the first aspect. The non-uniformity correction module is used to perform non-uniformity correction on the original input response of each pixel unit based on the calibration model of each pixel unit, so as to obtain the corrected output image.

[0017] Fifthly, this application provides a computer program product, including a computer program or instructions that, when run on an electronic device, cause the electronic device to perform the method described in the first aspect above, or any possible implementation of the first aspect, or the method described in the second aspect above.

[0018] In summary, compared with the prior art, the technical solutions conceived in this application have the following main technical advantages: This application provides a method and system for non-uniformity calibration and correction of quantum dot shortwave infrared detectors. Unlike traditional linear correction methods that calculate gain and bias based solely on two points (dark and bright fields), this method addresses the non-linear differences in pixel response curves and inconsistencies in response amplitude and slope between different pixels in quantum dot shortwave infrared images under different operating conditions. It constructs a pixel-by-pixel polynomial response model and combines it with an operating condition axis interpolation mechanism to achieve multi-condition adaptive correction. The core idea is to first perform pixel-by-pixel polynomial calibration at multiple preset calibration anchor points, and then, during runtime, interpolate the coefficients of the polynomial calibration model corresponding to adjacent anchor points based on the current operating condition to obtain a calibration model that matches the current operating condition, thereby completing the non-uniformity correction of the input image.

[0019] This application provides a method and system for non-uniformity calibration and correction of quantum dot short-wave infrared detectors, particularly suitable for quantum dot short-wave infrared detectors. Since the response characteristics of quantum dot short-wave infrared detectors are affected not only by the readout circuit but also by the quantum dot photosensitive film, interface state, carrier transport characteristics, dark current, and temperature-dependent characteristics, their non-uniformity is not simply a fixed gain bias error, but often manifests as nonlinear response differences that vary with brightness, integration time, gain, and focal plane temperature. To address these characteristics, this application employs a pixel-by-pixel polynomial calibration model to fit the pixel response curve and generates a calibration model for the current operating condition through a multi-condition interpolation mechanism. This method can uniformly express the non-uniformity in quantum dot short-wave infrared detectors caused by material response differences, pixel response nonlinearity, and operating condition changes as a pixel-by-pixel parameter model, thereby improving the correction stability in low-light, long integration time, high-gain, and temperature-varying scenarios. Compared with traditional two-point calibration, this application not only reduces fixed pattern noise and residual fringes but also improves the local brightness mismatch and image style drift problems caused by operating condition switching in quantum dot short-wave infrared images. In particular, for the spatial variation noise, dark field shadows and pixel response discreteness commonly found in quantum dot short-wave infrared detectors, this application improves the adaptability of the calibration model to the actual device response by using multi-brightness point fitting, pixel-by-pixel normalization and operating axis interpolation. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the non-uniformity calibration method for quantum dot shortwave infrared detectors provided in this application embodiment; Figure 2 This is a flowchart illustrating the non-uniformity correction method for quantum dot shortwave infrared detectors provided in this application embodiment; Figure 3 This is a schematic diagram comparing two-point polynomial non-uniformity correction provided in the embodiments of this application; Figure 4 This is a schematic diagram of the interpolation effect of the integral time condition provided in the embodiments of this application; Figure 5 This is an architecture diagram of the non-uniformity calibration system for a quantum dot short-wave infrared detector provided in an embodiment of this application; Figure 6 This is an architecture diagram of the non-uniformity correction system for a quantum dot short-wave infrared detector provided in an embodiment of this application; Figure 7 This is an architectural diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0023] Furthermore, throughout this specification, references to "an embodiment"; "an embodiment," "an example," or similar language indicate that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of this application. Therefore, the appearance of the phrase "in one embodiment;" throughout this specification, and similar language, may, but not necessarily, refer to the same embodiment.

[0024] Figure 1 This is a flowchart illustrating the non-uniformity calibration method for a quantum dot short-wave infrared detector provided in this application embodiment; the aforementioned quantum dot short-wave infrared detector includes multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity in the original detected image; such as Figure 1 As shown, the method includes the following steps: Step S101: Determine multiple pre-set calibration conditions; the detector conditions include at least one of integration time, gain, and focal plane temperature. Step S102: Under each calibration condition, obtain the different original input response values ​​of each pixel unit to light signal input with different preset light intensities; and determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. Step S103: Under each calibration condition, a calibration function is fitted between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals to determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response so that the corrected input response matches the ideal input response value.

[0025] In a preferred embodiment, the detector is a quantum dot short-wave infrared detector, specifically a short-wave infrared detector based on a photosensitive layer formed from PbS, HgTe, InAs, or other colloidal quantum dot materials. The quantum dot short-wave infrared detector can be integrated with a silicon-based readout circuit to form an area array imaging device. Compared with traditional InGaAs short-wave infrared detectors, this type of device has advantages such as low fabrication cost, easily scalable pixel size and array scale, and tunable spectral response. However, due to differences in material distribution, film deposition, interface states, and charge transport in the quantum dot photosensitive layer, its pixel response non-uniformity and condition-dependent variations are more complex. Public information also indicates that quantum dot SWIR image sensors are considered a promising alternative for constructing low-cost SWIR image sensors.

[0026] In one possible implementation, the method further includes: Determine the current operating status of the quantum dot shortwave infrared detector; From the plurality of calibration conditions, determine at least two calibration conditions that are adjacent to the current calibration condition's parameters; Based on the interval of operating parameters between the current operating condition and each adjacent calibration operating condition, the calibration model of the corresponding pixel unit under each adjacent calibration operating condition is interpolated to obtain the calibration model of each pixel unit under the current operating condition.

[0027] In one possible implementation, when the detector's operating conditions include only one of integration time, gain, and focal plane temperature, linear interpolation is performed on the calibration models of corresponding pixel units under two adjacent calibration conditions to obtain the calibration models of each pixel unit under the current operating condition. And / or when the quantum dot shortwave infrared detector operating conditions include at least two of integration time, gain, and focal plane temperature, multidimensional interpolation is performed on the calibration models of corresponding pixel units under each adjacent calibration condition based on the relative positions of the operating parameters of the current operating condition on the operating condition axes corresponding to different operating conditions, so as to obtain the calibration model of each pixel unit under the current operating condition.

[0028] In one possible implementation, the calibration function is a polynomial function; and / or the basis functions of the polynomial model are Legendre basis functions.

[0029] Taking the case where only integration time is included as an example, this application does not directly establish a unified mapping for the entire image, but instead establishes a response curve model for each pixel separately. Specifically, a set of polynomial models is established for each pixel within a calibration grid composed of multiple integration time anchor points and multiple brightness calibration points (i.e., light signals with preset light intensities). During runtime, temporary coefficients are obtained by interpolating between adjacent anchor points based on the current integration time. The input pixel is then normalized and substituted into this temporary polynomial to obtain the corrected output. The key to this scheme is not "simplely switching a certain parameter," but rather "continuously generating the correction parameters for the current pixel along the case axis."

[0030] In one possible implementation, fitting the calibration function includes: The original input response value of each pixel unit to different light signals is normalized to obtain the normalized original input response value; The ideal input response value of each pixel unit to different light signals is normalized to obtain the normalized ideal input response value; Using the normalized original input response value as the input value and the normalized ideal input response value as the output value, the polynomial function is fitted to determine the coefficients of each order of polynomial, thus obtaining the corresponding calibration model.

[0031] In one possible implementation, determining the ideal input response value for all pixel units to each preset light intensity light signal input includes: Acquire T detection images of the detector for the same preset light intensity light signal input, where T is a positive integer greater than 1; Determine the mean input response for each detected image; The ideal input response value of the preset light intensity signal is obtained by averaging the T input responses corresponding to the T detection images.

[0032] In one embodiment, to unify subsequent notation, the pixel set, the integral time anchor point set, and the brightness calibration point set are defined as follows: 1. Pixel index set 2. Set of integration time anchor points 3. Set of brightness calibration points in, and These represent the image height and width, respectively. A set of pixel locations These represent the vertical and horizontal coordinates of a pixel, respectively. For the set of time anchor points for integration; This is the set of brightness calibration points.

[0033] 4. Define the set of calibration sample indices. as follows: For any pixel Its complete set of calibration observations can be represented as: in, Indicates the time anchor point of integration. and brightness calibration point Below, pixels The calibration observations (i.e., the original input response) are used. Multiple frames of original images can be acquired for each grid point, and the mean frame or median frame is taken as the representative image to reduce the impact of random noise and occasional outliers on the fitting results.

[0034] 5. Pixel-by-pixel input normalization and target brightness normalization For any pixel First, the minimum and maximum response values ​​of the pixel are statistically analyzed across all calibration samples. Then, the input of the pixel is compressed to... A unified domain is defined. The purpose of this is to first eliminate dynamic range differences between pixels, allowing subsequent polynomials to primarily learn "curve shape correction" rather than simultaneously undertaking amplitude scaling. At the same time, the calibrated brightness is mapped to a unified target domain to ensure consistency in the fitted target.

[0035] 5.1 pixel minimum response value 5.2 pixel maximum response value The minimum and maximum response values ​​mentioned above are statistical results for a single pixel p across all calibrated samples, rather than global statistics for the entire image.

[0036] 5.3 pixel dynamic range scale In numerical implementation, when When the value is too small, introduce a positive number. Implement numerical stability protection.

[0037] 5.4 Define pixel-wise normalized mapping as follows: Where x represents the pixel input response value to be normalized, which can be an observation value from the calibration phase. It can also be the original detector output response value of pixel p in the input image under the current operating conditions during runtime.

[0038] Therefore, at the calibration point Below, pixels The normalized input can be represented as: 5.5 Define the upper and lower bounds of brightness normalization: Further define the target brightness normalization mapping : Pixel-wise input normalization mapping Mapping with target brightness normalization The difference in their definition stems from the fact that they apply to different objects: (1) Input normalization mapping A normalization interval is constructed for each pixel, and its normalization interval is determined by the minimum and maximum response values ​​of that pixel in all calibration samples, which is used to eliminate the dynamic range difference between pixels. (2) Target brightness normalization mapping A globally unified mapping, whose normalization interval is determined by a calibrated brightness set, is used to define a unified fitting target space.

[0039] The above design eliminates the amplitude differences between pixels on the input side during model fitting, while maintaining a unified reference standard on the output side, thereby improving fitting stability and cross-pixel consistency.

[0040] 5.6 Therefore, the first The normalized target value corresponding to each brightness calibration point is: 5.7 Pixel-by-pixel polynomial modeling For fixed pixels and fixed integral time anchor Define the pixel response correction function: use The expansion of the first-order basis functions is as follows: in, For a pre-selected set of basis functions, For pixels At the integration time anchor point The next Model coefficients; This is the corrected input response value, which is the ideal input response value for each preset light intensity optical signal input.

[0041] If a standard power basis is used, then we can take: The model can now be written as: when When, it can be written in third-order form: If a Legendre orthogonal basis is used, then variable transformation is first introduced: The first four Legendre polynomials can be expressed as: .

[0042] Correspondingly, the following can be taken: .

[0043] Using Legendre orthogonal bases has the following advantages: (1) Reduce the numerical ill-conditioned nature of polynomial fitting In high-order polynomial fitting, ordinary power-order basis functions can easily lead to a large condition number in the design matrix, resulting in numerical instability or even amplifying noise errors when solving least-squares problems. Legendre basis functions, due to their orthogonality between orders, can significantly reduce matrix correlation and improve solution stability.

[0044] (2) Improve the separability of the fitting coefficients Under an orthogonal basis, coefficients of each order correspond to independent components of different orders, which reduces the coupling between model parameters and is beneficial for subsequent parameter interpolation and regularization control.

[0045] (3) Suppressing higher-order oscillations Ordinary power polynomials are prone to the Runge phenomenon near the interval boundaries, where the polynomial oscillates violently around the two ends of the interval, causing the interpolation result to deviate significantly from the original function. In contrast, Legendre polynomials have a more uniform approximation property, reducing higher-order oscillations while maintaining fitting accuracy, thus improving the smoothness of the non-uniformity correction results.

[0046] (4) It is beneficial to the generalization ability of the pixel-by-pixel model. In pixel-by-pixel modeling scenarios, the response curves of different pixels vary greatly. Using orthogonal basis functions can maintain good fitting consistency under a uniform model order, thereby improving the stability of the overall correction effect.

[0047] 5.8 Solving for polynomial coefficients For each pixel unit, the raw input response value and the corresponding ideal input response value to different optical signals (i.e., The calibration function between the two is fitted to determine the calibration model corresponding to each pixel unit.

[0048] The methods for solving the above polynomial coefficients can be found in relevant prior art records, and will not be described in detail in the embodiments of this application.

[0049] Figure 2 This is a schematic flowchart of the non-uniformity correction method for quantum dot short-wave infrared detectors provided in this application embodiment; as shown... Figure 2 As shown, it includes the following steps: Step S201: Determine the current operating condition of the detector and obtain the original input response of each pixel unit under the current operating condition; Step S202: Using the calibration method provided in the above embodiment, obtain the calibration model of each pixel unit under the current working condition; Step S202: Based on the calibration model of each pixel unit, perform non-uniformity correction on the original input response of each pixel unit to obtain the corrected output image.

[0050] It should be noted that this scheme employs a pixel-by-pixel modeling method at the algorithm level for non-uniformity correction of detectors such as short-wave infrared detectors. This method differs from traditional linear correction methods that calculate gain and bias based solely on two points, dark and bright fields. Instead, it addresses the non-linear differences in pixel response curves of quantum dot short-wave infrared images at different integration times, as well as the inconsistencies in response amplitude and slope between different pixels. It constructs a pixel-by-pixel polynomial response model and combines it with a working condition axis interpolation mechanism to achieve multi-condition adaptive correction. The core idea is to first perform pixel-by-pixel polynomial calibration at multiple preset integration time anchor points, and then, during runtime, interpolate the polynomial coefficients corresponding to adjacent anchor points based on the current working integration time to obtain a temporary correction model matching the current working condition, thereby completing the non-uniformity correction of the input image.

[0051] In the current embodiment, the operating condition axis is the integration time. Without changing the overall modeling framework, the operating condition axis can also be extended to a focal plane temperature axis, a gain axis, or a combined operating condition axis composed of integration time, temperature, and gain. This extension method is suitable for multi-segment calibration and parameter calling in more complex environments, but does not change the overall technical approach of this solution: "pixel-by-pixel modeling + operating condition axis interpolation + parameterized deployment".

[0052] For the same pixel Multiple sets of coefficient vectors can be obtained at different integration time anchor points. During runtime, when the current integration time... When the value falls between two adjacent anchor points, the interpolation weights are calculated first, and then each order of coefficients is interpolated to obtain the current temporary coefficient vector. It should be emphasized that the object of interpolation in this scheme is not the final output image, nor a simple gain-bias table, but the parameter vector of the pixel-wise polynomial model.

[0053] The system can establish multiple sets of calibration parameters according to preset operating condition combinations; the operating conditions include at least integration time, and may also include gain mode and focal plane temperature. During operation, based on the current working state, the system first determines the operating condition range, and then directly selects the closest parameter range or performs interpolation between adjacent anchor point parameters to obtain the current correction parameters.

[0054] The system can adaptively match the parameter level by comprehensively considering the current integration time, gain status, focal plane temperature, and image statistical features (global mean, histogram distribution, saturated pixel ratio, dark area ratio, local contrast, etc.).

[0055] Extended explanation: When the system simultaneously collects integration time, gain status and focal plane temperature, the determination of adjacent calibration anchor points in step S6 can be extended to locate two or more adjacent calibration anchor points in the multidimensional operating space and perform multidimensional interpolation on their vertex parameters.

[0056] Specifically, in terms of implementation, the multi-condition parameters can be organized and stored in non-volatile memory in the form of a lookup table, and thus can be understood as a multi-dimensional parameter LUT. When the condition dimension is three-dimensional, trilinear interpolation can be used, or the local cube can be divided into several simplexes and tetrahedral interpolation can be used. When the condition dimension is extended to four dimensions or above, multilinear interpolation is preferred, or the local hypercube can be divided into several high-dimensional simplexes and simplex interpolation can be performed.

[0057] It should be emphasized that the object of the multidimensional interpolation is not the original image grayscale values ​​themselves, but the parameter vector of the pixel-by-pixel correction model. Therefore, the essence of this scheme is not to perform direct 3D / 4D color lookup table mapping on the image in the traditional sense, but to interpolate the parameter field under discrete anchor points under multiple working conditions to generate a temporary correction model adapted to the current working conditions.

[0058] Preferably, the operating condition vector is defined as follows, where integration time, gain, and focal plane temperature are representative variables of the operating condition dimension. In the current preferred embodiment, the actual implementation can be regarded as a degradation of the above-mentioned multidimensional operating condition interpolation on a one-dimensional integration time axis; when the operating condition dimension is extended from one dimension to two, three, or higher dimensions, its mathematical form remains consistent, with only the number of vertices and the corresponding weight calculation method of the local operating condition unit changing.

[0059] To further expand the applicability of the algorithm, the current single-axis interpolation based primarily on integration time can be extended to multi-dimensional parameter field interpolation under multiple discrete anchor points for various operating conditions. The basic idea is to set discrete calibration anchor points on multiple operating condition axes, such as integration time, gain state, and focal plane temperature, and perform pixel-by-pixel parameter calibration at each anchor point combination. During runtime, based on the relative position of the current operating condition on each operating condition axis, multi-dimensional interpolation is performed on the parameter vectors at the vertices of local operating condition units to generate temporary correction parameters corresponding to the current operating condition, thus obtaining the calibration model for the current operating condition.

[0060] In one specific embodiment, the above-described non-uniformity correction method may include the following steps: Step S1: Set multiple integration time anchor points according to the preset calibration scheme, and set multiple brightness calibration points under each integration time anchor point.

[0061] Step S2: Acquire shortwave infrared raw image sequences under each calibration condition, and calculate representative images for each condition. The representative images can be mean frames or median frames.

[0062] Step S3: Based on all calibrated samples, calculate the minimum and maximum response values ​​for each pixel and establish pixel-by-pixel normalization parameters.

[0063] Step S4: At each integration time anchor point, fit the brightness response polynomial for each pixel to generate pixel-by-pixel polynomial coefficients.

[0064] Step S5: Write the pixel-by-pixel polynomial coefficients and normalization parameters into a non-volatile memory at a predetermined address for use by the main control chip or FPGA during runtime.

[0065] Step S6: When the module is running, the current integration time is obtained, and the adjacent calibration anchor points corresponding to the current integration time are determined in the preferred implementation method; when the system considers both the gain state and the focal plane temperature, the local working condition unit where the current working condition is located is located in the multi-dimensional working condition space.

[0066] Step S7: Calculate the interpolation weights based on the relative positions of the current working condition on each working condition axis; for the single-axis case, perform one-dimensional interpolation on the pixel-by-pixel polynomial coefficients of adjacent anchor points; for the multi-working-condition case, perform multi-dimensional interpolation on the pixel-by-pixel parameter vectors at the vertices of local working condition units (i.e., two or more adjacent working condition anchor points).

[0067] Step S8: Perform non-uniformity correction on the current input original image using the pixel-by-pixel polynomial model obtained by interpolation.

[0068] Step S9: Send the corrected image to the subsequent bad pixel processing, noise reduction, sharpening and dynamic range adjustment modules, and output it synchronously through dual video interfaces.

[0069] Understandably, if the system also collects gain status and focal plane temperature, then before step S6, the operating condition grouping can be determined based on the gain mode and focal plane temperature, and then parameter selection or multi-axis interpolation can be performed.

[0070] This application employs a pixel-wise polynomial modeling approach, which can more accurately describe the nonlinear response relationship of pixels under different brightness and integration time conditions, improving the accuracy of non-uniformity correction. A parameterized modeling and runtime interpolation mechanism based on integration time anchor points is used, allowing correction parameters to change continuously with integration time, reducing image jumps and fixed pattern residuals during switching between different exposure conditions. A pixel-wise normalization modeling approach reduces the impact of differences in the dynamic range of different pixels on fitting stability, improving model robustness. Orthogonal basis functions, robust loss, and regularization strategies are used to suppress the influence of abnormal calibration points and higher-order oscillations on the fitting results, improving calibration stability. Furthermore, the interpolation mechanism of the multi-condition parameter field enables this scheme to maintain a unified parameter calling framework when integration time, gain, and temperature change, facilitating a smooth expansion from the current single-axis implementation to a multi-axis implementation, and also benefiting the modular storage of parameter tables and real-time access on the FPGA side.

[0071] Figure 3 The image shows a comparison of the correction effects of the two-point algorithm and the polynomial correction algorithm in this invention on the same short-wave infrared image data. The comparison chart shows the polynomial correction and two-point correction: on the scatter plot of the same pixel response curve, both the two-point linear fitting curve and the polynomial fitting curve are plotted simultaneously to illustrate that the proposed solution has a stronger fitting ability for nonlinear responses. Figure 3 The two-point correction results still show some residual vertical stripes of fixed noise, and the polynomial correction effect is significantly better for the chamfering problem caused by process residue.

[0072] Figure 4 This is a comparison chart of the polynomial correction effect before and after interpolation under the same extreme data and different integration time conditions. It can be seen that the polynomial coefficients running for 15ms with an integration time of 1ms are completely unable to correct the non-uniformity of the extreme case, while the polynomial coefficients after using the interpolation method can completely cover such extreme cases.

[0073] Figure 5 This is an architectural diagram of the non-uniformity calibration system for a quantum dot short-wave infrared detector provided in an embodiment of this application; as shown... Figure 5 As shown, it includes: The calibration condition determination module 510 is used to determine multiple preset calibration conditions; the detector's conditions include at least one of integration time, gain, and focal plane temperature. The input response determination module 520 is used to acquire different original input response values ​​of each pixel unit to light signal input with different preset light intensities under each calibration condition; and to determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. The calibration model determination module 530 is used to fit a calibration function between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals under each calibration condition, and determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response so that the corrected input response matches the ideal input response value.

[0074] It should be understood that the above system is used to execute the methods in the above embodiments. The corresponding program modules in the system are similar in implementation principle and technical effect to those described in the above methods. The working process of the system can be referred to the corresponding process in the above methods, and will not be repeated here.

[0075] Figure 5 The non-uniformity calibration system shown here is only an example of the above-mentioned functional module division when processing data. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the non-uniformity calibration system can be divided into different functional modules to complete all or part of the functions described above.

[0076] Figure 6 This is an architectural diagram of the non-uniformity correction system for a quantum dot short-wave infrared detector provided in an embodiment of this application; as shown... Figure 6 As shown, it includes: The raw response determination module 610 is used to determine the current operating condition of the detector and to obtain the raw input response of each pixel unit under the current operating condition. The calibration model determination module 620 is used to obtain the calibration model of each pixel unit under the current working condition using the calibration method described in the above embodiments. The non-uniformity correction module 630 is used to perform non-uniformity correction on the original input response of each pixel unit based on the calibration model of each pixel unit, so as to obtain the corrected output image.

[0077] It should be understood that the above system is used to execute the methods in the above embodiments. The corresponding program modules in the system are similar in implementation principle and technical effect to those described in the above methods. The working process of the system can be referred to the corresponding process in the above methods, and will not be repeated here.

[0078] Figure 6The non-uniformity correction system shown is only illustrated by the division of the above functional modules when processing data. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the non-uniformity correction system can be divided into different functional modules to complete all or part of the functions described above.

[0079] It should be noted that this application discloses a non-uniformity calibration and correction method and system for quantum dot short-wave infrared detectors. This method addresses the condition-related non-uniformity in quantum dot short-wave infrared detectors caused by material differences in the quantum dot photosensitive layer, dark current, spatial variation noise, pixel response nonlinearity, and variations in integration time, gain, and focal plane temperature. Under multiple calibration conditions, the method acquires the original input response values ​​of each pixel unit to different preset light intensities, and fits a pixel-by-pixel polynomial calibration model based on the original input response values ​​and the ideal input response values. During operation, the calibration models under adjacent calibration conditions are interpolated according to the current condition to obtain the pixel correction model under the current condition, and the original image is then corrected for non-uniformity. This application improves the accuracy and stability of non-uniformity correction for quantum dot short-wave infrared images in low-light, long integration time, high-gain, and temperature-varying scenarios.

[0080] Based on the methods in the above embodiments, this application provides an electronic device, such as... Figure 7 As shown, the electronic device may include a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 730, the communication interface 720, and the memory 710 communicate with each other via the communication bus 740. The processor 710 can call logical instructions in the memory to execute the methods in the above embodiments.

[0081] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0082] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0083] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0084] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0085] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0086] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0087] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.

[0088] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for non-uniformity calibration of a quantum dot short-wave infrared detector, wherein the quantum dot short-wave infrared detector comprises multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity of the detected original image, characterized in that... The methods include: Determine multiple pre-set calibration conditions; The detector's operating conditions include at least one of the following: integration time, gain, and focal plane temperature; Under each calibration condition, acquire the different original input response values ​​of each pixel unit to light signal input with different preset light intensities; and determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. Under each calibration condition, a calibration function is fitted between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals to determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response so that the corrected input response matches the ideal input response value.

2. The detector non-uniformity calibration method according to claim 1, characterized in that, Also includes: Determine the current operating status of the detector; From the plurality of calibration conditions, determine at least two calibration conditions that are adjacent to the current calibration condition's parameters; Based on the interval of operating parameters between the current operating condition and each adjacent calibration operating condition, the calibration model of the corresponding pixel unit under each adjacent calibration operating condition is interpolated to obtain the calibration model of each pixel unit under the current operating condition.

3. The method according to claim 2, characterized in that, When the detector's operating conditions include only one of integration time, gain, and focal plane temperature, linear interpolation is performed on the calibration models of corresponding pixel units under two adjacent calibration conditions to obtain the calibration models of each pixel unit under the current operating condition. And / or when the detector operating condition includes at least two of integration time, gain, and focal plane temperature, multidimensional interpolation is performed on the calibration models of corresponding pixel units under each adjacent calibration condition based on the relative positions of the operating parameters of the current operating condition on the operating axis corresponding to different operating conditions, so as to obtain the calibration model of each pixel unit under the current operating condition.

4. The method according to any one of claims 1 to 3, characterized in that, The calibration function is a polynomial function; and / or the basis functions of the polynomial model are Legendre basis functions.

5. The method according to claim 4, characterized in that, Fitting the calibration function includes: The original input response value of each pixel unit to different light signals is normalized to obtain the normalized original input response value; The ideal input response value of each pixel unit to different light signals is normalized to obtain the normalized ideal input response value; Using the normalized original input response value as the input value and the normalized ideal input response value as the output value, the polynomial function is fitted to determine the coefficients of each order of polynomial, thus obtaining the corresponding calibration model.

6. The method according to claim 1, characterized in that, Determine the ideal input response value of all pixel units to a light signal input of a preset light intensity, including: Acquire T detection images of the detector for the same preset light intensity light signal input, where T is a positive integer greater than 1; Determine the mean input response for each detected image; The ideal input response value of the preset light intensity signal is obtained by averaging the T input responses corresponding to the T detection images.

7. A method for non-uniformity correction of a quantum dot short-wave infrared detector, wherein the quantum dot short-wave infrared detector comprises multiple pixel units, and the responses of different pixel units to the same input differ, resulting in non-uniformity of the original image detected by the detector, characterized in that... The methods include: Determine the current operating condition of the detector and obtain the raw input response of each pixel unit under the current operating condition; Using the calibration method described in any one of claims 1 to 6, obtain the calibration model of each pixel unit under the current working condition; Based on the calibration model of each pixel unit, the original input response of each pixel unit is non-uniformity corrected to obtain the corrected output image.

8. A non-uniformity calibration system for a quantum dot short-wave infrared detector, wherein the quantum dot short-wave infrared detector comprises multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity of the detected original image, characterized in that... system include: The calibration condition determination module is used to determine multiple pre-set calibration conditions; the detector's conditions include at least one of the following: integration time, gain, and focal plane temperature. The input response determination module is used to acquire the different original input response values ​​of each pixel unit to light signal input with different preset light intensities under each calibration condition; and to determine the ideal input response values ​​of all pixel units to light signal input with each preset light intensity. The calibration model determination module is used to fit a calibration function between the original input response value and the corresponding ideal input response value of each pixel unit for different optical signals under each calibration condition, and determine the calibration model corresponding to each pixel unit; the calibration model is used to correct the original input response value of the pixel unit to obtain the corrected input response, so that the corrected input response matches the ideal input response value.

9. A non-uniformity correction system for a quantum dot short-wave infrared detector, wherein the quantum dot short-wave infrared detector comprises multiple pixel units, and the responses of different pixel units to the same optical signal input differ, resulting in non-uniformity of the detected original image, characterized in that... The system includes: The raw response determination module is used to determine the current operating condition of the detector and to obtain the raw input response of each pixel unit under the current operating condition. The calibration model determination module is used to obtain the calibration model of each pixel unit under the current working condition using the calibration method described in any one of claims 1 to 6. The non-uniformity correction module is used to perform non-uniformity correction on the original input response of each pixel unit based on the calibration model of each pixel unit, so as to obtain the corrected output image.

10. A computer program product, comprising a computer program or instructions, characterized in that: When the computer program or instructions are run on an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1 to 6 or claim 7.

Images