Apparatus for acquiring a mathematical model for eliminating residual image in infrared images and method of using the same

Abstract

The application discloses a device for acquiring a mathematical model for eliminating residual image of infrared image and a use method thereof, which comprises a black body, a lens, a baffle, a signal processing board and an interface board connected in sequence; the lens is directed to the black body and collects infrared rays emitted and / or reflected by the black body; the baffle is an infrared shutter; the signal processing board is used for controlling the lens and the baffle to work according to instructions from a computer; and the interface board is used for connecting the signal processing board with the computer through a network port, sending data to the computer and accepting instructions from the computer. In use, data acquisition is firstly performed; then a decay function is fitted; and finally the fitted decay function is verified, which specifically comprises multiple experimental verification, chip working temperature influence verification and adaptability verification at different temperatures. The application can better adapt to residual image characteristics, increase accuracy, reduce noise interference on fitting, enhance robustness, reduce redundant calculation, and effectively improve processing speed on the premise of ensuring residual image elimination effect.

Description

Technical Field

[0001] This invention relates to the field of infrared image processing technology, and in particular to an apparatus and a method for using it to obtain a mathematical model for eliminating infrared image ghosting. Background Technology

[0002] The sensor's response to temperature changes and its processing speed often cause image ghosting in infrared sensors.

[0003] When a sensor detects a temperature change, especially when an object is moving rapidly or the ambient temperature fluctuates significantly, the sensor's response time may not be synchronized with the actual change, resulting in image information from the previous frame remaining in the new frame.

[0004] In addition, the latency and insufficient processing power of image processing algorithms may also cause the information of the previous state in the image to fail to be updated in time, resulting in ghosting.

[0005] Other heat sources in the environment, such as sunlight, the heat effect of the sensor itself, or the heat dissipation of the equipment, can also interfere with the sensor's measurement results and exacerbate the generation of image ghosting.

[0006] The presence of ghosting can affect the quality of subsequently acquired images and the quality of the final displayed image.

[0007] If an infrared sensor exhibits ghosting, it cannot be eliminated directly through hardware; it requires natural attenuation based on the attenuation rate of different sensors.

[0008] Existing image retention methods all fit a function to the entire image as a whole, which is often limited by the global features of the image.

[0009] Therefore, how to better adapt to the characteristics of ghosting, increase accuracy, reduce the interference of noise on fitting, enhance robustness, and at the same time reduce redundant calculations, while ensuring the effect of ghosting elimination and effectively improving the processing speed, has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0010] In view of the above-mentioned deficiencies of the prior art, the present invention provides an apparatus and a method for obtaining a mathematical model for eliminating infrared image ghosting. The purpose is to better adapt to ghosting characteristics, increase accuracy, reduce noise interference on fitting, enhance robustness, and reduce redundant calculations, thereby effectively improving processing speed while ensuring the ghosting elimination effect.

[0011] To achieve the above objectives, the present invention discloses an apparatus for obtaining a mathematical model for eliminating infrared image ghosting, comprising a blackbody, and a lens, a baffle, a signal processing board and an interface board connected in sequence.

[0012] The lens is directed toward the blackbody to collect infrared light emitted and / or reflected by the blackbody;

[0013] The baffle is an infrared shutter;

[0014] The signal processing board is used to control the operation of the lens and the baffle according to the instructions issued by the computer;

[0015] The interface board is used to connect the signal processing board to the computer via its built-in network port, send data to the computer, and receive instructions from the computer.

[0016] Preferably, the lens is a dual-lens lens; the baffle is an XL-SU-172T3 type infrared baffle.

[0017] Preferably, the computer uses Python software to control the raising and lowering of the baffle and saves the infrared image data captured by the lens.

[0018] The present invention also provides a method for using an apparatus for obtaining a mathematical model for eliminating infrared image ghosting, comprising the following steps:

[0019] Step 1: Data Acquisition;

[0020] Step 2: Fit the decay function;

[0021] Step 3: Verify the fitted decay function, which includes multiple experimental verifications, verification of the effect of chip operating temperature, and verification of adaptability to different temperatures.

[0022] Preferably, step 1 is as follows:

[0023] Step 1.1: Clear the function set cache of the computer's Python software to ensure that there are no functions in the function set cache;

[0024] Step 1.2: Adjust the blackbody temperature to 20°C, and wait for the blackbody temperature to stabilize at 20°C. Then, use the Python software to collect 10 frames of bright field image data and use the Python software to read all pixels of all the bright field image data to determine whether the bright field image data is stable.

[0025] If the bright field image data is unstable, wait for the blackbody temperature to stabilize at 20°C, then re-acquire 10 frames of the bright field image data and make the judgment again until the 10 frames of the bright field image data are stable.

[0026] Step 1.3: Send a descent command to the baffle using the Python software to make the baffle fall completely, and then save some dark field image data;

[0027] All dark field image data that have undergone bad pixel correction and non-uniformity correction, and have no other noise except for afterimages and have a uniform background are saved;

[0028] Step 1.4: Send an upward or downward command to the baffle using the Python software to make the baffle fully rise.

[0029] More preferably, step 2 is as follows:

[0030] Step 2.1: Find the number of dark-field images in the data of all the dark-field images where the first afterimage disappears, and use the number of dark-field images where the first afterimage disappears and all the dark-field image data before it as the afterimage generation to dissipation process;

[0031] Step 2.2: Divide the pixels of all i frames of dark field image data in the process of afterimage generation to dissipation into nine parts in a 3x3 grid pattern, and take two pixels at the center of each grid of the dark field image data. Save all the pixels to obtain 18×i pixel data.

[0032] Step 2.3: Based on the natural decay law of image afterimage, fit the pixel data belonging to the same 3x3 grid to obtain 18 first decay functions with respect to time; each 3x3 grid has two of the first decay functions;

[0033] Step 2.4, normalization processing, as follows:

[0034] Step 2.4.1: Determine whether the two first decay functions belonging to the same 3x3 grid can be normalized;

[0035] If all two first attenuation functions belonging to the same 3x3 grid can be normalized, then all two first attenuation functions belonging to the same 3x3 grid are normalized to obtain 9 second attenuation functions, and then step 2.4.A2 is executed.

[0036] If any two of the first attenuation functions belonging to the same 9x9 grid cannot be normalized, then each 9x9 grid is split into 9x9 grids again to form an 81x8 grid. Two points are taken from the center of each small grid of the 81x8 grid corresponding to the pixel of each dark field image data to save, resulting in 162×i 81x8 grid pixels, and then step 2.4.B2 is executed.

[0037] Step 2.4.A2: Determine whether the nine second decay functions can be normalized;

[0038] If the nine second decay functions can be normalized, then the nine second decay functions are normalized to obtain the third decay function, and the third decay function is output as the decay function to complete the fitting.

[0039] If the nine second decay functions cannot be normalized, then determine whether the two second decay functions belonging to two adjacent nine-square grids can be normalized;

[0040] If both of the second decay functions belonging to any two adjacent 3x3 grids can be normalized, then normalize all the two second decay functions belonging to two adjacent 3x3 grids, and finally fit x fourth decay functions; where 1 <x<9;

[0041] And output the fourth decay function described in x as the decay function to complete the fitting;

[0042] If neither of the two second decay functions belonging to two adjacent 3x3 grids can be normalized, then the nine second decay functions will be output as the decay functions to complete the fitting.

[0043] Step 2.4.B2: For points belonging to the same small cell in the 81-grid, according to the natural decay law of image afterimage, fit them sequentially to obtain 162 sixth decay functions of the 81-grid with respect to time; wherein, each small cell in the 81-grid has two of the sixth decay functions;

[0044] Step 2.4.B3: Determine whether the two sixth decay functions belonging to each small cell in the 81-grid can be normalized;

[0045] If any two of the sixth decay functions belonging to each small cell in the 81-grid cannot be normalized, then all steps are re-executed starting from step 1.

[0046] If all two sixth decay functions belonging to each small cell in the 81-grid can be normalized, then the two sixth decay functions belonging to each small cell in the 81-grid are normalized to form 81 seventh decay functions, and the subsequent steps are continued.

[0047] Step 2.4.B4: Determine whether the nine seventh decay functions belonging to each 3x3 grid can be normalized;

[0048] If the nine seventh decay functions belonging to each nine-square grid can be normalized, then the normalization forms nine eighth decay functions as the decay function output to complete the fitting.

[0049] If the nine seventh attenuation functions belonging to any one of the nine-square grids cannot be normalized, determine whether the nine seventh attenuation functions can be normalized with the nine seventh attenuation functions of an adjacent another nine-square grid;

[0050] If the nine seventh attenuation functions can be normalized with the nine seventh attenuation functions of an adjacent another nine-square grid, normalize the nine seventh attenuation functions with the nine seventh attenuation functions of the adjacent another nine-square grid, and finally fit y attenuation functions, where 9 < y < 8; and output the y eighth attenuation functions as the attenuation functions that have completed fitting;

[0051] Otherwise, output the 81 seventh attenuation functions as the attenuation functions that have completed fitting.

[0052] More preferably, the multiple experimental verifications are specifically as follows:

[0053] Step 3.A1: Send a baffle lowering instruction through the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a baffle raising instruction through the Python software to make the baffle rise;

[0054] Step 3.A2: Randomly select two consecutive frames of the dark field image data, and calculate the difference V between the theoretical and actual results through the following formula;

[0055] V = f z '-K(t)*f z ' -1 ;

[0056] where, K(t) is the fitted attenuation function in Step 2; f z ' is the second frame of the dark field image data in the two frames of the dark field image data; f z ' -1 is the first frame of the dark field image data in the two frames of the dark field image data;

[0057] Step 3.A3: Compare the difference V between the theoretical and actual results with a threshold; if the difference V between the theoretical and actual results is within the threshold, the attenuation function fitting is correct; otherwise, start from Step 1 and re-execute all steps.

[0058] More preferably, the verification of the influence of the chip operating temperature is specifically as follows:

[0059] Step 3.B1: Replace the blackbody with the chip for the working time T, send a baffle lowering instruction through the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a baffle raising instruction through the Python software to make the baffle rise;

[0060] Step 3.B2: Randomly select two consecutive frames of the dark field image data, and calculate the difference V between the theoretical and actual results using the following formula;

[0061] V = f z '-K(t)*f z ' -1 ;

[0062] Where K(t) is the fitting decay function in step 2; f z 'The second frame of the dark field image data in the two frames of dark field image data; f z ' -1 The first frame of the dark field image data in the two frames of dark field image data;

[0063] Step 3.B3: Compare the difference V between the theoretical and actual results with a threshold. If the difference V is within the threshold, it means that the chip temperature has no effect on the decay function. Otherwise, it means that the chip temperature has an effect on the decay function, and all steps need to be executed again from step 1 after time T.

[0064] Step 4.B3: Save the function corresponding to 20°C for the obtained chip.

[0065] More preferably, the adaptability verification at different temperatures is specifically as follows:

[0066] Step 3.C1: Adjust the blackbody temperature to 60°C and wait for the blackbody temperature to stabilize at 60°C;

[0067] Step 3.C2: Send a command to lower the baffle using the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a command to raise the baffle using the Python software to make the baffle rise.

[0068] Step 3.C3: Perform step 2 to fit the decay function of the blackbody at 60℃;

[0069] Step 3.C4: Perform the multiple experimental verifications and the chip operating temperature effect verification to ensure the accuracy of the decay function at the current temperature, and save the obtained decay function at the current temperature;

[0070] Step 3.C5: Compare the decay function of the blackbody at 60℃ and the decay function at 20℃ to determine whether they are consistent;

[0071] If they are consistent, it means that different temperatures have no effect on the function. The decay functions of the blackbody at 60℃ and 20℃ are then combined into a function of one temperature, and the adaptability verification of different temperatures is completed.

[0072] If there is a discrepancy, a baffle descent command is sent via the Python software to make the baffle fall completely. Then, the blackbody temperature is adjusted to 40°C. Step 2 is executed to obtain the blackbody attenuation function at 40°C. The multiple experimental verifications and the chip operating temperature influence verification are performed to ensure the accuracy of the attenuation function at the current temperature. The obtained attenuation function at the current temperature is then saved.

[0073] Step 3.C6: Compare the decay function of the blackbody at 40℃ and the decay function at 20℃ to determine whether they are consistent;

[0074] If they are consistent, then fit a set of decay functions every 20℃ and save the functions for all temperatures;

[0075] If they are inconsistent, a set of decay functions is fitted every 10℃, and the functions for all temperatures are saved.

[0076] More preferably, after completing step 3, the blackbody is replaced with the object whose infrared image ghosting needs to be eliminated, and the temperature of the object is measured using a thermometer; then, a command to lower the baffle is sent using Python to make the baffle fall completely, and an image is captured; next, a command to raise the baffle is sent using Python to raise the baffle, an image is captured, and the image is saved according to time; finally, the attenuation function verified in step 3 is called, and the ghosting elimination operation is performed by substituting it into the formula, as follows:

[0077] f m '=f m -K(t)*f m-1

[0078] Where K(t) is the decay function verified in step 3, f m It is the original image that was captured, f m The image after ghosting is removed is the output image.

[0079] The beneficial effects of this invention are:

[0080] The application of this invention enables the elimination of infrared image ghosting to better adapt to ghosting characteristics, increase accuracy, reduce noise interference with fitting, enhance robustness, and reduce redundant calculations, thereby effectively improving processing speed while ensuring the ghosting elimination effect.

[0081] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0082] Figure 1 A schematic diagram of an embodiment of the present invention is shown.

[0083] Figure 2 The flowchart of obtaining the mathematical model for eliminating infrared image ghosting is shown in one embodiment of the present invention.

[0084] Figure 3 A flowchart of fitting the decay function is shown in one embodiment of the present invention. Detailed Implementation

[0085] Example

[0086] like Figure 1 As shown, the device for obtaining a mathematical model for eliminating infrared image ghosting includes a blackbody, and a lens, a baffle, a signal processing board and an interface board connected in sequence.

[0087] The lens is pointed at the blackbody to collect infrared light emitted and / or reflected by the blackbody;

[0088] The baffle is an infrared shutter;

[0089] The signal processing board is used to control the operation of the lens and the baffle according to the instructions issued by the computer;

[0090] The interface board is used to connect the signal processing board to the computer via its built-in network port, sending data to the computer and receiving instructions from the computer.

[0091] In some embodiments, the lens is a bi-light lens; the baffle is an XL-SU-172T3 type infrared baffle.

[0092] Infrared baffles are a mature product. In this invention, the XL-SU-172T3 infrared baffle produced by Foshan Xieliang Optoelectronic Products Co., Ltd. is used. The red wire provides DC 4V voltage, and the black wire is GND.

[0093] The baffle is supplied with voltage through a pin connected to the FPGA board. When a 4V voltage is applied, the baffle falls down, and when the voltage disappears, the baffle rises up, completing the switching action.

[0094] In practical applications, the switch command for raising and lowering the baffle is embedded in the FPGA board, so that the corresponding instructions issued by the computer can interact with the FPGA board to complete the baffle switching operation.

[0095] In some embodiments, the computer uses Python software to control the raising and lowering of the baffle and saves the infrared image data captured by the lens.

[0096] like Figure 2 As shown, the present invention also provides a method for using an apparatus for obtaining a mathematical model for eliminating infrared image ghosting, comprising the following steps:

[0097] Step 1: Data Acquisition;

[0098] Step 2: Fit the decay function;

[0099] Step 3: Verify the fitted decay function, which includes multiple experimental verifications, verification of the effect of chip operating temperature, and verification of adaptability to different temperatures.

[0100] In some embodiments, step 1 is specifically as follows:

[0101] Step 1.1: Clear the function cache of your computer's Python software to ensure that there are no functions in the function cache;

[0102] Step 1.2: Adjust the blackbody temperature to 20℃ and wait for the blackbody temperature to stabilize at 20℃. Then, use Python software to collect 10 frames of bright field image data and read all pixels of all bright field image data to determine whether the bright field image data is stable.

[0103] If the bright field image data is unstable, wait for the blackbody temperature to stabilize at 20°C, then re-acquire 10 frames of bright field image data and make the judgment again until the 10 frames of bright field image data are stable.

[0104] Step 1.3: Send a descent command to the baffle using Python software to make the baffle fall completely, and then save some dark field image data;

[0105] All dark field image data that have undergone bad pixel correction and non-uniformity correction, and have no other noise except for afterimages and have a uniform background are saved;

[0106] Step 1.4: Send an up / down command to the baffle using Python software to raise the baffle completely.

[0107] like Figure 3 As shown, in some embodiments, step 2 is specifically as follows:

[0108] Step 2.1: Find the number of dark-field images in the saved dark-field image data at the moment when the first ghost image disappears. Take the number of dark-field images at the moment when the first ghost image disappears and all the dark-field image data before it as the process from ghost image generation to dissipation. In practical applications, all the dark-field image data before the number of dark-field images at the moment when the first ghost image disappears, sorted in chronological order, is the complete process from ghost image generation to dissipation.

[0109] Step 2.2: Divide all the pixels of the i-frame dark field image data in the process of afterimage generation to dissipation into nine parts in a nine-grid pattern, and take two pixels at the center of each grid of the nine-grid of each dark field image data. Save all the pixels to get 18×i pixel data.

[0110] Step 2.3: Based on the natural decay law of image afterimages, fit the pixel data belonging to the same 3x3 grid to obtain 18 first decay functions with respect to time; each 3x3 grid has two first decay functions;

[0111] Step 2.4, normalization processing, as follows:

[0112] Step 2.4.1: Determine whether the two first decay functions belonging to the same 3x3 grid can be normalized;

[0113] If all two first attenuation functions belonging to the same 3x3 grid can be normalized, then all two first attenuation functions belonging to the same 3x3 grid are normalized to obtain 9 second attenuation functions, and then step 2.4.A2 is executed.

[0114] If any two first attenuation functions belonging to the same 9x9 grid cannot be normalized, then each 9x9 grid is split into 9x9 grids again to form an 81x8 grid. Two points are taken from the center of each small grid of the 81x8 grid corresponding to the pixel of each dark field image data to save, resulting in 162×i 81x8 grid pixels, and then step 2.4.B2 is executed.

[0115] Step 2.4.A2: Determine whether the nine second decay functions can be normalized;

[0116] If the nine second decay functions can be normalized, then the nine second decay functions are normalized to obtain the third decay function, and the third decay function is output as the decay function to complete the fitting.

[0117] If the nine second decay functions cannot be normalized, then determine whether the two second decay functions belonging to two adjacent nine-square grids can be normalized.

[0118] If any two second decay functions belonging to two adjacent 3x3 grids can be normalized, then normalize all two second decay functions belonging to two adjacent 3x3 grids to finally fit x fourth decay functions; where 1 <x<9;

[0119] And output x fourth decay functions as the decay functions to complete the fitting;

[0120] If neither of the two second decay functions belonging to two adjacent 3x3 grids can be normalized, then the 9 second decay functions will be output as the decay functions to complete the fitting.

[0121] Step 2.4.B2: For points belonging to the same small cell in the 81-grid, according to the natural decay law of image afterimage, fit them sequentially to obtain 162 sixth decay functions of the 81-grid with respect to time; among them, each small cell in the 81-grid has two sixth decay functions.

[0122] Step 2.4.B3. Determine whether the two sixth attenuation functions belonging to each small cell in the 81 - cell grid can be normalized;

[0123] If the two sixth attenuation functions belonging to any small cell in the 81 - cell grid cannot be normalized, then start from Step 1 and execute all steps again;

[0124] If the two sixth attenuation functions belonging to all small cells in the 81 - cell grid can be normalized, then normalize the two sixth attenuation functions belonging to each small cell in the 81 - cell grid to form 81 seventh attenuation functions, and continue with the subsequent steps;

[0125] Step 2.4.B4. Determine whether the 9 seventh attenuation functions belonging to each nine - cell grid can be normalized;

[0126] If the 9 seventh attenuation functions belonging to each nine - cell grid can be normalized, then normalize them to form 9 eighth attenuation functions as the attenuation functions for completed fitting and output;

[0127] If the 9 seventh attenuation functions belonging to any nine - cell grid cannot be normalized, then determine whether the 9 seventh attenuation functions can be normalized with the 9 seventh attenuation functions of an adjacent nine - cell grid;

[0128] If the 9 seventh attenuation functions can be normalized with the 9 seventh attenuation functions of an adjacent nine - cell grid, then normalize the 9 seventh attenuation functions with the 9 seventh attenuation functions of the adjacent nine - cell grid, and finally fit out y attenuation functions, where 9 < y < 8; and output the y eighth attenuation functions as the attenuation functions for completed fitting;

[0129] Otherwise, output the 81 seventh attenuation functions as the attenuation functions for completed fitting.

[0130] In some embodiments, the multiple - experiment verification is as follows:

[0131] Step 3.A1. Send a baffle - lowering instruction through Python software to make the baffle fully drop, continuously save i - frame dark - field image data, and then send a baffle - raising instruction through Python software to make the baffle rise;

[0132] Step 3.A2. Randomly select two consecutive frames of dark - field image data, and calculate the theoretical and actual result difference V through the following formula:

[0133] V = f z '- K(t)*f z ' -1 ;

[0134] where, K(t) is the fitting attenuation function of Step 2; fz 'This refers to the second dark-field image data in a two-frame dark-field image dataset; f z ' -1 This refers to the first dark-field image data in a two-frame dark-field image dataset.

[0135] This invention obtains the second dark-field image data from the theoretical two-frame dark-field image data by multiplying the first frame dark-field image data by the fitting attenuation function in step 2. The difference between the second frame dark-field image data actually captured and the first frame dark-field image data is then calculated. The closer the result is to 0, the better the attenuation function is fitted.

[0136] Step 3.A3: Compare the difference V between the theoretical and actual results with the threshold; if the difference V is within the threshold, the attenuation function is correctly fitted; otherwise, repeat all steps from step 1.

[0137] In some embodiments, the verification of the effect of chip operating temperature is as follows:

[0138] Step 3.B1: Replace the blackbody with the chip and work for time T. Send a baffle descent command through Python software to make the baffle fall completely. Continuously save i frames of dark field image data. Then send a baffle ellipse command through Python software to make the baffle rise.

[0139] Step 3.B2: Randomly select two consecutive frames of dark field image data, and calculate the difference V between the theoretical and actual results using the following formula;

[0140] V = f z '-K(t)*f z ' -1 ;

[0141] Where K(t) is the fitting decay function in step 2; f z 'This refers to the second dark-field image data in a two-frame dark-field image dataset; f z ' -1 This refers to the first dark-field image data in a two-frame dark-field image dataset.

[0142] Step 3.B3: Compare the difference V between the theoretical and actual results with the threshold. If the difference V is within the threshold, it means that the chip temperature has no effect on the decay function. Otherwise, it means that the chip temperature has an effect on the decay function, and all steps need to be executed again from step 1 after time T.

[0143] Step 4.B3: Save the function corresponding to 20℃ for the obtained chip.

[0144] In some embodiments, the adaptability verification at different temperatures is specifically as follows:

[0145] Step 3.C1: Adjust the blackbody temperature to 60℃ and wait for the blackbody temperature to stabilize at 60℃;

[0146] Step 3.C2: Send a command to lower the baffle using Python software to make the baffle fall completely, continuously save i frames of dark field image data, and then send a command to raise the baffle using Python software to make the baffle rise.

[0147] Step 3.C3: Perform step 2 to fit the blackbody's decay function at 60℃;

[0148] Step 3.C4: Perform multiple experimental verifications and chip operating temperature effect verifications to ensure the accuracy of the decay function at the current temperature, and save the obtained decay function at the current temperature;

[0149] Step 3.C5: Compare the blackbody's decay function at 60℃ and at 20℃ to determine if they are consistent;

[0150] If they are consistent, it means that different temperatures have no effect on the function. The decay functions of the blackbody at 60℃ and 20℃ are then combined into a function of one temperature, and the fitness verification at different temperatures is completed.

[0151] If there is a discrepancy, a command to lower the baffle is sent via Python software to make the baffle fall completely. Then, the blackbody temperature is adjusted to 40℃, and step 2 is executed to obtain the blackbody attenuation function at 40℃. Multiple experiments and verifications of the effect of chip operating temperature are performed to ensure the accuracy of the attenuation function at the current temperature. The obtained attenuation function at the current temperature is then saved.

[0152] Step 3.C6: Compare the blackbody's decay function at 40℃ and at 20℃ to determine if they are consistent;

[0153] If they are consistent, then fit a set of decay functions every 20℃ and save the functions for all temperatures;

[0154] If they are inconsistent, a set of decay functions is fitted every 10℃, and the functions for all temperatures are saved.

[0155] This invention obtains all functions by conducting multiple experimental verifications, chip characteristic verifications, and temperature adaptability verifications on the fitted decay function. The above-mentioned technical means have the following effects:

[0156] First, multiple experiments can identify differences in the intensity and disappearance speed of different afterimages, making the fitting function consistent in different scenarios and reducing errors.

[0157] Secondly, the infrared sensor chip's temperature may rise during long-term operation, potentially affecting its image retention characteristics. This verification ensures that the fitting function remains accurate even with rising chip temperature, avoiding image retention changes caused by temperature accumulation, thereby improving the stability of the calibration method under long-term operating conditions.

[0158] Then, different blackbody temperatures were used to replace high and low temperature scenarios. This verification ensures that the fitted decay function can still effectively eliminate image retention under different ambient temperatures, guaranteeing its stability and improving its adaptability.

[0159] Finally, all verified functions are saved to the function set cache for easy access in subsequent practical applications.

[0160] The above-mentioned technical means distinguish this invention from existing methods and highlight its accuracy, reliability and robustness in image retention correction.

[0161] In some embodiments, after completing step 3, the blackbody is replaced with the object whose infrared image ghosting needs to be eliminated, and the temperature of the object is measured using a thermometer. Then, a command to lower the baffle is sent using Python to make the baffle fall completely, and an image is captured. Next, a command to raise the baffle is sent using Python to make the baffle rise, an image is captured, and the image is saved according to time. Finally, the attenuation function verified in step 3 is called, and the ghosting elimination operation is performed using the following formula:

[0162] f m '=f m -K(t)*f m-1

[0163] Where K(t) is the decay function verified in step 3, f m It is the original image that was captured, f m The image after ghosting is removed is the output image.

[0164] This invention automatically calls the corresponding image attenuation function in the function set buffer based on the current temperature and time after obtaining each image frame. This function removes image ghosting from the image of that frame. The output image of the current frame is obtained by subtracting the attenuated image of the previous frame from the image of the second frame; this is also the image without image ghosting. This invention does not modify the original image in any way; it simply multiplies the corresponding attenuation value from the previous frame to obtain the image ghosting value. Subtracting the calculated image ghosting value from the original image yields the final output image without image ghosting.

[0165] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. An apparatus for acquiring a mathematical model for eliminating infrared image ghosting; characterized in that, This includes a blackbody, and in sequence, a lens, a baffle, a signal processing board, and an interface board; The lens is directed toward the blackbody to collect infrared light emitted and / or reflected by the blackbody; The baffle is an infrared shutter; The signal processing board is used to control the operation of the lens and the baffle according to the instructions issued by the computer; The interface board is used to connect the signal processing board to the computer via its built-in network port, send data to the computer, and receive instructions from the computer.

2. The apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 1, characterized in that, The lens is a dual-lens transmission lens; the baffle is an XL-SU-172T3 type infrared baffle.

3. The apparatus and method for obtaining a mathematical model for eliminating infrared image ghosting according to claim 1, characterized in that, The computer uses Python software to control the raising and lowering of the baffle and saves the infrared image data captured by the lens.

4. A method for using an apparatus for obtaining a mathematical model for eliminating infrared image ghosting, characterized in that, Includes the following steps: Step 1: Data Acquisition; Step 2: Fit the decay function; Step 3: Verify the fitted decay function, which includes multiple experimental verifications, verification of the effect of chip operating temperature, and verification of adaptability to different temperatures.

5. A method for using an apparatus for obtaining a mathematical model for eliminating infrared image ghosting, characterized in that, Step 1 is as follows: Step 1.1: Clear the function set cache of the computer's Python software to ensure that there are no functions in the function set cache; Step 1.2: Adjust the blackbody temperature to 20°C, and wait for the blackbody temperature to stabilize at 20°C. Then, use the Python software to collect 10 frames of bright field image data and use the Python software to read all pixels of all the bright field image data to determine whether the bright field image data is stable. If the bright field image data is unstable, wait for the blackbody temperature to stabilize at 20°C, then re-acquire 10 frames of the bright field image data and make the judgment again until the 10 frames of the bright field image data are stable. Step 1.3: Send a descent command to the baffle using the Python software to make the baffle fall completely, and then save some dark field image data; All dark field image data that have undergone bad pixel correction and non-uniformity correction, and have no other noise except for afterimages and have a uniform background are saved; Step 1.4: Send an upward or downward command to the baffle using the Python software to make the baffle fully rise.

6. The method of using the apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 5, characterized in that, Step 2 is as follows: Step 2.1: Find the number of dark-field images in the data of all the dark-field images where the first afterimage disappears, and use the number of dark-field images where the first afterimage disappears and all the dark-field image data before it as the afterimage generation to dissipation process; Step 2.2: Divide the pixels of all i frames of dark field image data in the process of afterimage generation to dissipation into nine parts in a 3x3 grid pattern, and take two pixels at the center of each grid of the dark field image data. Save all the pixels to obtain 18×i pixel data. Step 2.3: Based on the natural decay law of image afterimage, fit the pixel data belonging to the same 3x3 grid to obtain 18 first decay functions with respect to time; each 3x3 grid has two of the first decay functions; Step 2.4, normalization processing, as follows: Step 2.4.1: Determine whether the two first decay functions belonging to the same 3x3 grid can be normalized; If all two first attenuation functions belonging to the same 3x3 grid can be normalized, then all two first attenuation functions belonging to the same 3x3 grid are normalized to obtain 9 second attenuation functions, and then step 2.4.A2 is executed. If any two of the first attenuation functions belonging to the same 9x9 grid cannot be normalized, then each 9x9 grid is split into 9x9 grids again to form an 81x8 grid. Two points are taken from the center of each small grid of the 81x8 grid corresponding to the pixel of each dark field image data to save, resulting in 162×i 81x8 grid pixels, and then step 2.4.B2 is executed. Step 2.4.A2: Determine whether the nine second decay functions can be normalized; If the nine second decay functions can be normalized, then the nine second decay functions are normalized to obtain the third decay function, and the third decay function is output as the decay function to complete the fitting. If the nine second decay functions cannot be normalized, then determine whether the two second decay functions belonging to two adjacent nine-square grids can be normalized; If both of the second decay functions belonging to any two adjacent 3x3 grids can be normalized, then normalize all the two second decay functions belonging to two adjacent 3x3 grids, and finally fit x fourth decay functions; where 1 <x<9; And output the fourth decay function described in x as the decay function to complete the fitting; If neither of the two second decay functions belonging to two adjacent 3x3 grids can be normalized, then the nine second decay functions will be output as the decay functions to complete the fitting. Step 2.4.B2: For points belonging to the same small cell in the 81-grid, according to the natural decay law of image afterimage, fit them sequentially to obtain 162 sixth decay functions of the 81-grid with respect to time; wherein, each small cell in the 81-grid has two of the sixth decay functions; Step 2.4.B3: Determine whether the two sixth decay functions belonging to each small cell in the 81-grid can be normalized; If any two of the sixth decay functions belonging to each small cell in the 81-grid cannot be normalized, then all steps are re-executed starting from step 1. If all two sixth decay functions belonging to each small cell in the 81-grid can be normalized, then the two sixth decay functions belonging to each small cell in the 81-grid are normalized to form 81 seventh decay functions, and the subsequent steps are continued. Step 2.4.B4: Determine whether the nine seventh decay functions belonging to each 3x3 grid can be normalized; If the nine seventh decay functions belonging to each nine-square grid can be normalized, then the normalization forms nine eighth decay functions as the decay function output to complete the fitting. If the nine seventh decay functions belonging to any 3x3 grid cannot be normalized, then determine whether the nine seventh decay functions can be normalized to the nine seventh decay functions of the adjacent 3x3 grid. If the nine seventh attenuation functions described in item 9 can be normalized with the nine seventh attenuation functions of an adjacent another nine-square grid, then the nine seventh attenuation functions are normalized with the nine seventh attenuation functions of the adjacent another nine-square grid, and finally y attenuation functions are fitted, where 9 < y < 8; and the y eighth attenuation functions are output as the attenuation functions that have completed fitting; On the contrary, the 81 seventh attenuation functions are output as the attenuation functions that have completed fitting.

7. The method of using the apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 6, characterized in that, The specific multiple experimental verifications are as follows: Step 3.A1: Send a baffle lowering instruction through the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a baffle raising instruction through the Python software to make the baffle rise; Step 3.A2: Randomly select two consecutive frames of the dark field image data, and calculate the theoretical and actual result difference V through the following formula; V=f z '-K(t)*f z ' -1 ; Where K(t) is the fitting decay function in step 2; f z 'The second frame of the dark field image data in the two frames of dark field image data; f z ' -1 The first frame of the dark field image data in the two frames of dark field image data; Step 3.A3: Compare the theoretical and actual result difference V with the threshold; if the theoretical and actual result difference V is within the threshold, the attenuation function fitting is correct; otherwise, start from step 1 and re-execute all steps.

8. The method of using the apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 6, characterized in that, The verification of the influence of the chip working temperature is as follows: Step 3.B1: Replace the black body with the chip for working time T, send a baffle lowering instruction through the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a baffle raising instruction through the Python software to make the baffle rise; Step 3.B2: Randomly select two consecutive frames of the dark field image data, and calculate the theoretical and actual result difference V through the following formula; V=f z '-K(t)*f z ' -1 ; Where K(t) is the fitting decay function in step 2; f z 'The second frame of the dark field image data in the two frames of dark field image data; f z ' -1 The first frame of the dark field image data in the two frames of dark field image data; Step 3.B3: Compare the theoretical and actual result difference V with the threshold; if the theoretical and actual result difference V is within the threshold, it means that the chip temperature has no influence on the attenuation function; otherwise, it means that the chip temperature has an influence on the attenuation function, and then it is necessary to start from step 1 again after T time and re-execute all steps; Step 4.B3: Save the function corresponding to the chip at 20°C obtained.

9. The method of using the apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 6, characterized in that, The verification of the adaptability at different temperatures is as follows: Step 3.C1: Adjust the temperature of the black body to 60°C and wait for the temperature of the black body to stabilize at 60°C; Step 3.C2: Send a baffle lowering instruction through the Python software to make the baffle fall completely, continuously save i frames of the dark field image data, and then send a baffle raising instruction through the Python software to make the baffle rise; Step 3.C3: Execute step 2 to fit the attenuation function of the black body at 60°C; Step 3.C4: Execute the multiple experimental verifications and the verification of the influence of the chip working temperature to ensure the accuracy of the attenuation function at the current temperature, and save the attenuation function at the current temperature obtained; Step 3.C5: Compare the attenuation function of the black body at 60°C with the attenuation function at 20°C to judge whether they are consistent; If they are consistent, it means that different temperatures have no effect on the function. The decay functions of the blackbody at 60℃ and 20℃ are then combined into a function of one temperature, and the adaptability verification of different temperatures is completed. If there is a discrepancy, a baffle descent command is sent via the Python software to make the baffle fall completely. Then, the blackbody temperature is adjusted to 40°C. Step 2 is executed to obtain the blackbody attenuation function at 40°C. The multiple experimental verifications and the chip operating temperature influence verification are performed to ensure the accuracy of the attenuation function at the current temperature. The obtained attenuation function at the current temperature is then saved. Step 3.C6: Compare the decay function of the blackbody at 40℃ and the decay function at 20℃ to determine whether they are consistent; If they are consistent, then fit a set of decay functions every 20℃ and save the functions for all temperatures; If they are inconsistent, a set of decay functions is fitted every 10℃, and the functions for all temperatures are saved.

10. The method of using the apparatus for obtaining a mathematical model for eliminating infrared image ghosting according to claim 4, characterized in that, After completing step 3, replace the blackbody with the object whose infrared image ghosting needs to be eliminated, and measure the temperature of the object using a thermometer. Then, use Python to send a command to lower the baffle completely, and take an image. Next, use Python to send a command to raise the baffle, and take an image, saving it according to time. Finally, call the attenuation function verified in step 3, and substitute it into the formula to perform the ghosting elimination operation. The formula is as follows: f m '=f m -K(t)*f m-1 Where K(t) is the decay function verified in step 3, f m It is the original image captured, f m The image after ghosting is removed is the output image.

Images