Adaptive calibration method and system for multi-stage gain image detector in laboratory

Abstract

The application provides a kind of adaptive scaling method and system of laboratory multi-gain image detector, belongs to the field of image processing and detector scaling technology, detector has N gain, method includes: respectively in each gain position acquisition dark field image and multilevel radiance bright field image;For each pixel under each gain position, based on the multilevel radiance response data fitted response function collected, gain coefficient and bias term of each pixel under each position are obtained;Select reference gain position, calculate the pixel level equivalent gain ratio and bias compensation of other positions relative to reference position, establish cross-position mapping lookup table;Gain independent correction model is constructed, the original image collected under any gain position is mapped to uniform response reference, and the corrected image is output.The application can realize full-position coverage, independently calibrate each gain position of detector, establish complete pixel level response model, and solve the pain point of re-calibration after multi-position switching.

Description

Technical Field

[0001] This invention belongs to the field of image processing and detector calibration technology, specifically relating to an adaptive calibration method and system for a laboratory multi-gain image detector. Background Technology

[0002] In laboratory scientific imaging (such as fluorescence microscopy, low-temperature physics observation, and high-speed imaging analysis), a single fixed gain mode often fails to meet the acquisition requirements of a wide dynamic range: weak signal scenes require high gain to improve the signal-to-noise ratio, while strong signal scenes require low gain to prevent saturation. Therefore, modern scientific image sensors (such as CMOS and CCD) are typically designed with multiple programmable gain levels for users to choose from, with typical configurations including 2, 4, or more levels. However, the introduction of multiple gain levels brings new calibration challenges. First, the detector response curves differ significantly at different gain levels—low gain results in a higher dynamic range but higher noise, while high gain increases sensitivity but reduces the dynamic range, making it difficult to observe multiple targets. Second, there is a lack of traceable mapping between the gain levels. After the user manually switches the gain, the image brightness jumps, and consistency cannot be restored by simple linear scaling, severely affecting subsequent quantitative analysis (such as light intensity measurement and concentration inversion). In existing technologies, although automatic gain control (AGC) methods can achieve automatic gain adjustment, they usually perform real-time feedback within a continuously adjustable gain range, failing to solve the pre-calibration problem of discrete multi-level gains. While radiometric calibration methods for multi-detector mosaic cameras involve calibration at multiple radiance levels, they primarily address consistency correction between different detectors (such as the ultraviolet imager on the Haiyang-1 C / D satellite), rather than response calibration between different gain levels of the same detector. Furthermore, flat-field correction methods typically assume a fixed gain; when gain levels are switched, the original correction parameters (dark-field offset, bright-field gain coefficient) become inapplicable, leading to residual non-uniformity in the calibrated image. Therefore, a method is urgently needed that can independently calibrate each gain level of the detector, establish response mapping relationships between levels, and achieve one-click calibration after gain switching to meet the high-precision imaging requirements of laboratories. Summary of the Invention

[0003] To overcome the shortcomings of existing laboratory multi-gain detectors, such as inconsistent responses between different gain levels, the need for recalibration after switching, and the lack of a unified mapping relationship, an adaptive calibration method is provided that can calibrate each gain level separately and establish the correlation between levels, ensuring the consistency of image brightness and response when switching between different gain levels.

[0004] The technical concept of this invention is as follows:

[0005] A two-step strategy of "independent calibration at different gain levels" and "cross-gain synthesis mapping" is employed to establish the response function for each gain level and construct a gain-independent correction model with the intermediate gain level as a reference. Specifically, bright-field and dark-field images are acquired separately for each gain level under a standard light source, and the pixel response slope and bias are calculated for each level. Then, using a reference level (such as the intermediate gain) as a reference, the equivalent gain ratio and bias compensation for other levels are calculated, forming a calibration lookup table. In practical use, when the user switches gain levels, the system automatically calls the corresponding correction parameters, achieving seamless calibration that is "ready to use immediately." The specific technical solution is as follows:

[0006] An adaptive calibration method for a laboratory multi-gain image detector, the detector having N gain levels, the method comprising the following steps:

[0007] Step S1: Acquire dark field images and multi-level radiance bright field images at each gain level;

[0008] Step S2: For each pixel at each gain level, fit the response function based on the collected multi-level radiance response data to obtain the gain coefficient of each pixel at each gain level. and bias terms ;

[0009] Step S3: Select the reference gain level and calculate the pixel-level equivalent gain ratio of each of the other levels relative to the reference level. and bias compensation amount Establish a cross-file mapping lookup table;

[0010] Step S4: Construct a gain-independent correction model to map the original images acquired at any gain level to a unified response benchmark and output the corrected image.

[0011] An adaptive calibration system for a laboratory multi-gain image detector implementing the method, comprising:

[0012] Image sensor with multi-level programmable gain control;

[0013] Computer storage unit, used to store pixel-level response parameters and cross-gain mapping lookup tables for each gain level;

[0014] The correction processing unit, connected to the image sensor and the storage unit, is used to read the corresponding parameters according to the current gain level and perform pixel-by-pixel correction.

[0015] The gear detection unit is used to monitor the gain gear status in real time and apply parameter correction.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method.

[0017] The beneficial effects of this invention are as follows:

[0018] 1. Full Gain Coverage: Each gain level of the detector is independently calibrated to establish a complete pixel-level response model, solving the pain point of needing to recalibrate after switching between multiple gain levels.

[0019] 2. Cross-gain consistency: By establishing a mapping relationship between gain levels, a unified response benchmark for images under different gain levels is achieved, ensuring the consistency of brightness and grayscale values ​​before and after switching, and meeting the needs of quantitative analysis.

[0020] 3. High-precision calibration: Pixel-level response fitting is used, which can significantly improve the accuracy of non-uniformity calibration compared to traditional global calibration methods.

[0021] 4. Drift compensation: Supports periodic recalibration and parameter updates, effectively suppressing response drift of the detector during long-term use, and ensuring the timeliness and reliability of calibration. Attached Figure Description

[0022] Figure 1 This is an overall flowchart of the method of the present invention;

[0023] Figure 2 This is a schematic diagram of the pixel response curve fitting at each gain level in this invention;

[0024] Figure 3 This is a schematic diagram illustrating the establishment of cross-level mapping relationships in this invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other. To achieve the above objectives, this invention adopts the following technical solution.

[0026] This invention provides an adaptive calibration method and system for a laboratory multi-gain image detector, wherein the detector has N programmable gain levels (N≥4), and includes the following steps:

[0027] S1: Separate calibration for different data types – Dark field and bright field acquisition

[0028] Under laboratory darkroom conditions, with the light source turned off, the detector gain was sequentially set to the i-th level (i=1,2,...,N), and P dark-field images were acquired. Calculate the mean value matrix of dark field pixels at each level. , For a pixel, such as Figure 1 As shown.

[0029] Under standard uniform light source (integrating sphere) illumination, K different radiance levels are set according to the camera's quantization level. (k=1,2,...,K,K≥5), acquire the bright field image at the corresponding radiance level for each gain level i, and obtain the pixel response value. .

[0030] S2: Pixel-level response function fitting

[0031] For each gain level i and each pixel The radiance-response line was fitted using the least squares method:

[0032] ;

[0033] in, For the i-th pixel The gain coefficient (response slope). This is the dark field bias term (dark field shift). Record the fitting parameters for each full pixel at each resolution to form a calibration parameter library, such as... Figure 1 As shown.

[0034] S3: Establishment of inter-gear mapping relationship

[0035] Select a gain level as a reference level (usually an intermediate level, such as the r-th level, where r = floor(N / 2)). For each pixel... Calculate the equivalent gain ratio of other gears i relative to the reference gear r:

[0036] ;

[0037] And the offset compensation amount:

[0038] ;

[0039] Establish cross-file mapping lookup table ,like Figure 1 As shown.

[0040] S4: Gain-independent correction model

[0041] The corrected target image pixel values ​​should satisfy the following condition: regardless of the detector's gain setting, the output should have a consistent corrected grayscale value under the same radiance. The correction formula is:

[0042] ;

[0043] in, For the corrected image, This is the original image. This formula maps images of any gain level to the same response reference, such as... Figure 1 As shown, the corrected image has the same brightness-grayscale response relationship as the image directly acquired at the reference setting r, thus achieving a gain-independent quantitative image.

[0044] S5: Adaptive scaling and gain switching lock

[0045] When a user switches the gain level via software, the system automatically performs the following operations:

[0046] 1) Gain Level Recognition: Read the current gain level index ;

[0047] 2) Parameter loading: Read the calibration parameters for this gear from the storage unit. , And cross-level mapping parameters;

[0048] 3) Real-time correction: For each frame of image acquired subsequently, pixel-by-pixel processing is performed according to the correction formula in step S4;

[0049] 4) Calibration verification: Collect 100 frames of reference light source images for calibration, and calculate the average of the single frame images to verify whether the calibrated mean is within the preset threshold. If it exceeds the threshold, recalibration is required.

[0050] S6: Periodic Recalibration and Drift Compensation

[0051] Because the detector may experience response drift during long-term use, the system supports periodic recalibration. This is achieved by reacquiring dark field and single-point bright field images at the current setting, and updating the calibration using single-point or two-point correction methods. and And update the cross-level mapping table accordingly.

[0052] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] Example 1: Calibration of a Four-Gap CMOS Camera

[0054] This embodiment applies to a laboratory-grade four-gain CMOS scientific camera (gain levels: 1x, 5x, 25x, 125x), such as... Figure 3 As shown.

[0055] Step 1: Independent calibration by grade

[0056] Place the camera in a dark box and turn off all light sources. Sequentially set the gain to 1x, 5x, 25x, and 125x, continuously capturing 100 frames of dark-field images at each gain level. Calculate the average value to obtain the dark-field bias term. .

[0057] Enable the integrating sphere uniform light source and set five radiance levels (corresponding to 10%, 30%, 50%, 70%, and 90% of the camera's dynamic range). At each gain level, sequentially acquire bright-field images for each radiance level, averaging 100 frames per level to reduce noise, and then subtract the corresponding dark-field offset to obtain the pixel response curve for each gain level, as shown below. Figure 2 As shown.

[0058] Step 2: Response function fitting

[0059] The least squares method was used to perform linear fitting on each pixel to obtain the response slope at each level. Experimental data shows that the low gain setting (125x) has the largest dynamic range, but the noise at the low-end input is relatively large; the high gain setting (1x) has a significantly improved sensitivity.

[0060] Step 3: Cross-file mapping

[0061] A 5x zoom level was selected as the reference level (balancing sensitivity and linearity). The pixel-level gain ratio of each zoom level relative to the 5x level was calculated. It is worth noting that, due to the nonlinearity of the circuit design, there is a deviation of about 5%-10% between the nominal gain ratio (e.g., 25x / 5x=5) and the actual pixel-level gain ratio, which is why this invention requires precise pixel-level calibration.

[0062] Step 4: Real-time calibration and switching

[0063] During the experiment, the user first acquired a weak signal image using the 1x gain setting under low-light conditions. The system automatically applied the correction parameters for the 1x gain setting and output the corrected image. The user then switched to a high-light area, and the software automatically switched the gain to the 125x gain setting. The system loaded the correction parameters for the 125x gain setting in real time and mapped the image to the same response reference according to the correction formula. The switching process was smooth, and the user observed no significant jumps in image brightness. Furthermore, the inter-pixel non-uniformity was reduced to below 1% after correction.

[0064] Step 5: Drift Compensation

[0065] The system collects dark field data at weekly intervals. If the dark field signal changes compared to previous data, a rapid recalibration is automatically performed: only the dark field and a single bright field (50% radiance) are collected, and the response parameters are updated using a two-point method. After recalibration, the image brightness shows no significant jump, and the non-uniformity between pixels is reduced to below 1% after correction, effectively compensating for the drift.

[0066] The present invention further provides an adaptive calibration method and system for a laboratory multi-gain image detector implementing the above method, comprising:

[0067] Image sensor with multi-level programmable gain control;

[0068] Computer storage unit, used to store pixel-level response parameters and cross-gain mapping lookup tables for each gain level;

[0069] The correction processing unit, connected to the image sensor and the storage unit, is used to read the corresponding parameters according to the current gain level and perform pixel-by-pixel correction.

[0070] The gear detection unit is used to monitor the gain gear status in real time and apply parameter correction.

[0071] Furthermore, it also includes a calibration verification module, which is used to acquire a reference light source image and verify the correction accuracy after gain switching. If the accuracy exceeds the threshold, a recalibration prompt will be issued.

[0072] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

Claims

1. An adaptive calibration method for a laboratory multi-gain image detector, characterized in that, The detector has N gain levels, and the method includes the following steps: Step S1: Acquire dark field images and multi-level radiance bright field images at each gain level; Step S2: For each pixel at each gain level, fit the response function based on the collected multi-level radiance response data to obtain the gain coefficient of each pixel at each gain level. and bias terms ; Step S3: Select the reference gain level and calculate the pixel-level equivalent gain ratio of each of the other levels relative to the reference level. and bias compensation amount Establish a cross-file mapping lookup table; Step S4: Construct a gain-independent correction model to map the original images acquired at any gain level to a unified response benchmark and output the corrected image.

2. The method according to claim 1, characterized in that, The response function described in step S2 is fitted using a linear model, and the fitting method is the least squares method.

3. The method according to claim 1, characterized in that, In step S3, the reference gain setting is selected as either the intermediate setting or the setting with the best linearity.

4. The method according to claim 1, characterized in that, The correction formula for the gain-independent correction model described in step S4 is: ； in, For the corrected image, For the original image, and This represents the gain ratio and the bias compensation.

5. The method according to claim 1, characterized in that, It also includes step S5: when the detector gain level is detected to be switching, the current level is automatically identified and the corresponding correction parameters are loaded to perform real-time correction on the subsequently acquired images.

6. The method according to claim 1, characterized in that, It also includes step S6: periodically perform recalibration by acquiring dark field and bright field images at the current gear and updating the response parameters.

7. The method according to claim 1, characterized in that, The cross-level mapping lookup table is as follows: .

8. An adaptive calibration system for a laboratory multi-gain image detector implementing the method of any one of claims 1-7, characterized in that, include: Image sensor with multi-level programmable gain control; Computer storage unit, used to store pixel-level response parameters and cross-gain mapping lookup tables for each gain level; The correction processing unit, connected to the image sensor and the storage unit, is used to read the corresponding parameters according to the current gain level and perform pixel-by-pixel correction. The gear detection unit is used to monitor the gain gear status in real time and apply parameter correction.

9. The system according to claim 8, characterized in that, It also includes a calibration verification module, which is used to acquire a reference light source image and verify the correction accuracy after gain switching. If the accuracy exceeds the threshold, a recalibration prompt will be issued.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method of any one of claims 1-7.

Images