A multi-parameter abnormal calibration system and method for infrared thermal imager
By designing a multi-parameter abnormal calibration system in an infrared thermal imager, generating focal plane position and angle labels, analyzing user control behavior, and realizing adaptive control of software and hardware, it solves the problem of the diversity of hardware control methods of infrared thermal imager The problem of differentiation of imaging effects has achieved the optimal regulation effect of combining software and hardware.
Patent Information
- Application Number
- CN202410552357.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-07
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-05-07
AI Technical Summary
There are multiple choices of control methods for infrared thermal imagers in hardware, which makes it difficult for software to optimize based on unified index parameters, resulting in differentiation of imaging effects and making it difficult to achieve optimal regulation of software and hardware combination.
Design a multi-parameter abnormal calibration system, including operation log module, data quality analysis module, data processing module and parameter calibration module, and generate position labels and angle labels of the focal plane, record and analyze user control behavior information, and randomly group them , classification and evaluation, standardize and calibrate based on the evaluation results, and realize adaptive regulation combining software and hardware.
Through unified hardware limitations and unified index parameters with the largest range, the differentiation of software and hardware combined with regulation can be reduced, abnormal calibration of infrared thermal imagers can be achieved, hardware limitations can be compensated, and the optimal software adaptability can be achieved.
Smart Images

Figure CN118225254B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of infrared thermal imagers, and in particular to a multi-parameter abnormal calibration system and method applied to infrared thermal imagers. Background Art
[0002] The infrared thermal imager realizes the imaging function through the focal plane array. The software optimizes the sensing behavior of each detector unit in the focal plane array to achieve focal plane compensation or calibration.
[0003] Then, the limitations of the focal plane hardware make it difficult to optimize software compensation or calibration based on unified indicator parameters. In many cases, the software has achieved maximum optimization, but there are still multiple control options in the hardware. Summary of the invention
[0004] The purpose of the present invention is to provide a multi-parameter abnormal calibration system and method for infrared thermal imagers to solve the problems raised in the above background technology.
[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0006] A multi-parameter abnormal calibration system applied to infrared thermal imagers, the system comprises: an operation log module, a data quality analysis module, a data processing module and a parameter calibration module;
[0007] The operation log module generates a position label and an angle label of the focal plane based on the internal structure design of the infrared thermal imager, and records the control behavior information of the user end;
[0008] The data quality analysis module is used to randomly divide the control behavior information into multiple groups, each group forming a subset data cluster, and analyze the data quality to perform quality screening on the control behavior information;
[0009] The data processing module classifies the control behavior information based on the screening result;
[0010] The parameter calibration module is used to evaluate the control behavior information through the evaluation index, and calibrate the control behavior based on the evaluation result.
[0011] Further, the operation log module includes a cache library unit and a tag pair unit;
[0012] The cache unit is used to establish an operation log cache, which stores control behavior information each time the imaging effect is dynamically adjusted, and the control behavior information includes the position information and the angle information of the focal plane; based on the internal structure design of the infrared thermal imager, a unified form of the focal plane position label and angle label are respectively established, and the position label and the angle label respectively have unique label attributes;
[0013] The label pair unit is used to record the i-th position label as , the jth angle label is recorded as , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as ,and ,in, represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, and E represents the total amount of control behavior information when the imaging effect is dynamically adjusted for the kth time, The position label indicating the focal plane and focal plane angle labels , constitutes a label pair.
[0014] Further, the data quality analysis module includes a subset construction unit and a quality analysis unit;
[0015] The subset forming unit constructs a behavior representation two-dimensional coordinate system based on the control behavior information, wherein the horizontal coordinate of the behavior representation two-dimensional coordinate system corresponds to the position label of the focal plane, and the vertical coordinate of the behavior representation two-dimensional coordinate system corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, the control behavior information is punctuated in the behavior representation two-dimensional coordinate system; The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as ,and , , x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information As a subset of data clusters The centroids of the subset data clusters are obtained respectively The punctuation and control behavior information corresponding to each control behavior information The Euclidean distance between the corresponding punctuation points, based on the Euclidean distance, the subset data cluster is obtained The intra-cluster distance is denoted as ;
[0016] The quality analysis unit is used to construct a data quality iterative analysis model. In each iteration, Randomly select a sample input, and the sample input selected in each iteration is different;
[0017] Controlling behavioral information As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows:
[0018]
[0019] in, Indicates control behavior information The data quality value of represents the expected value of the minimum intra-cluster distance;
[0020] Preset data quality threshold. If the data quality value is greater than or equal to the data quality threshold, the control behavior information is extracted. And save it in the output information set Otherwise, the control behavior information is removed ;
[0021] Until the control behavior set After every control behavior information in participates in the iteration, the iteration stops.
[0022] Further, the data processing module includes a first classifier unit and a second classifier unit;
[0023] The first classifier unit is used to output the information set Perform the first behavior classification, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, recorded as ,in, Represents the position information of the focal plane is the first classification reference, corresponding to the first classification set generated;
[0024] The second classifier unit is used to output the information set Perform the second behavior classification, use the angle information of the same focal plane as the second classification reference, traverse the label pairs with the same second classification reference, and generate the second classification set, recorded as ,in, Represents the angle information of the focal plane It is the second classification reference, corresponding to the generated second classification set.
[0025] Furthermore, the parameter calibration module includes an indicator parameter unit and a calibration analysis unit;
[0026] The indicator parameter unit is used to establish an evaluation indicator parameter category information library, in which the evaluation indicator parameter category information for evaluating the regulation effect is stored; the evaluation indicator parameters are uniformly encoded to form an evaluation indicator parameter set with a unique encoding attribute, which is recorded as ,in, represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters;
[0027] The calibration analysis unit is used to evaluate the index parameter set Each evaluation index parameter controls the behavior information Conduct assessments to obtain control behavior information Based on the initial evaluation mean of each control behavior information in the first classification set, the position information is obtained The comprehensive evaluation mean is recorded as ; Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information is obtained The comprehensive evaluation mean is denoted as ;
[0028] Respectively and The corresponding position labels and angle labels are used to calibrate the position and angle of the focal plane.
[0029] A multi-parameter abnormal calibration method for an infrared thermal imager, the method comprising the following steps:
[0030] Step S100: Based on the internal structure design of the infrared thermal imager, generate a position label and an angle label of the focal plane, and record the control behavior information of the user end;
[0031] Step S200: randomizing the control behavior information into multiple groups, each group forming a subset data cluster, and analyzing the data quality to perform quality screening on the control behavior information;
[0032] Step S300: Classifying the control behavior information based on the screening result;
[0033] Step S400: Evaluate the control behavior information through evaluation indicators, and calibrate the control behavior based on the evaluation results.
[0034] Furthermore, the specific implementation process of step S100 includes:
[0035] Step S101: Establishing an operation log cache library, wherein the operation log cache library stores control behavior information each time the imaging effect is dynamically adjusted, wherein the control behavior information includes the position information of the focal plane and the angle information of the focal plane; based on the internal structure design of the infrared thermal imager, establishing a unified form of the position label and the angle label of the focal plane, wherein the position label and the angle label have unique label attributes respectively;
[0036] Step S102: Record the i-th position tag as , the jth angle label is recorded as , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as ,and ,in, represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, and E represents the total amount of control behavior information when the imaging effect is dynamically adjusted for the kth time, The position label indicating the focal plane and focal plane angle labels , constitutes a label pair.
[0037] Furthermore, the specific implementation process of step S200 includes:
[0038] Step S201: Based on the control behavior information, construct a behavior representation two-dimensional coordinate system, wherein the horizontal coordinate of the behavior representation two-dimensional coordinate system corresponds to the position label of the focal plane, and the vertical coordinate of the behavior representation two-dimensional coordinate system corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, punctuate the control behavior information in the behavior representation two-dimensional coordinate system; The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as ,and , , x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information As a subset of data clusters The centroids of the subset data clusters are obtained respectively The punctuation and control behavior information corresponding to each control behavior information The Euclidean distance between the corresponding punctuation points, based on the Euclidean distance, the subset data cluster is obtained The intra-cluster distance is denoted as ;
[0039] Step S202: Construct a data quality iterative analysis model. In each iteration, Randomly select a sample input, and the sample input selected in each iteration is different;
[0040] Controlling behavioral information As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows:
[0041]
[0042] in, Indicates control behavior information The data quality value of represents the expected value of the minimum intra-cluster distance;
[0043] Preset data quality threshold. If the data quality value is greater than or equal to the data quality threshold, the control behavior information is extracted. And save it in the output information set Otherwise, the control behavior information is removed ;
[0044] Until the control behavior set After every control behavior information in participates in the iteration, the iteration stops;
[0045] According to the above method, when using an infrared thermal imager during imaging, in terms of hardware, the changes in the focal plane at different positions and angles result in different imaging effects. In terms of software, different indicator parameters are used to optimize the hardware deficiencies. The data quality iterative analysis model uses the similarity analysis of intra-cluster distances to screen out hardware control methods for focal planes with similar characteristics. However, due to the limitations of hardware functions, it is not possible to present the same imaging effect by satisfying control behaviors with similar characteristics.
[0046] Furthermore, the specific implementation process of step S300 includes:
[0047] Step S301: Output information set To classify the behavior, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, recorded as ,in, Represents the position information of the focal plane is the first classification reference, corresponding to the first classification set generated;
[0048] Step S302: Using the angle information of the same focal plane as the second classification reference, traverse the label pairs with the same second classification reference, and generate a second classification set, recorded as ,in, Represents the angle information of the focal plane It is the second classification reference, corresponding to the generated second classification set.
[0049] Furthermore, the specific implementation process of step S400 includes:
[0050] Step S401: Establish an evaluation indicator parameter category information library, in which the evaluation indicator parameter category information for evaluating the regulation effect is stored; uniformly encode the evaluation indicator parameters to form an evaluation indicator parameter set with a unique encoding attribute, recorded as ,in, represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters;
[0051] Step S402: Evaluate the parameter set of the indicator Each evaluation index parameter controls the behavior information Conduct assessments to obtain control behavior information Based on the initial evaluation mean of each control behavior information in the first classification set, the position information is obtained The comprehensive evaluation mean is recorded as ; Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information is obtained The comprehensive evaluation mean is denoted as ;
[0052] Respectively and The corresponding position labels and angle labels are used to calibrate the position and angle of the focal plane;
[0053] According to the above method, in software, each evaluation index parameter participates in the evaluation of imaging effect, but the limitation of focal plane hardware makes the evaluation value always different. This differentiation is difficult to optimize through unified index parameters. Based on different states of the focal plane, the evaluation values of imaging effects of different index parameters are obtained respectively. The index parameters can be unified to the maximum extent based on unified hardware limitations, so that the differentiation of software and hardware combined regulation can be minimized as much as possible. and It represents different optimal performance evaluation values obtained under the unified indicator parameters of the maximum range formed by the unified position restriction and the unified angle restriction respectively; then, the position restriction and the angle restriction are combined to achieve the unification of hardware restrictions, and at the same time, the unified hardware restrictions are combined to achieve adaptive optimal control in software.
[0054] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: in a multi-parameter abnormal calibration system and method for infrared thermal imagers provided by the present invention, based on the internal structure design of the infrared thermal imager, the position label and angle label of the focal plane are generated, and the control behavior information of the user end is recorded; the control behavior information is randomly divided into multiple groups, each group constitutes a subset data cluster, and the data quality is analyzed to perform quality screening on the control behavior information; based on the screening results, the control behavior information is classified; the control behavior information is evaluated through evaluation indicators, and the control behavior is calibrated based on the evaluation results; thereby, it is possible to compensate for the limitations of hardware while realizing adaptive optimal control of software, and it is possible to unify the indicator parameters to the maximum extent based on unified hardware limitations, so that the differentiation of the combined control of software and hardware is minimized as much as possible, thereby realizing abnormal calibration of infrared thermal imagers. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0056] Figure 1 It is a structural schematic diagram of a multi-parameter abnormal calibration system applied to an infrared thermal imager of the present invention;
[0057] Figure 2 The present invention is a schematic diagram of the steps of a multi-parameter abnormal calibration method applied to an infrared thermal imager. DETAILED DESCRIPTION
[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0059] See also Figure 1-Figure 2 , the present invention provides a technical solution:
[0060] See also Figure 1 , in the first embodiment: a multi-parameter abnormal calibration system applied to an infrared thermal imager is provided, the system comprising: an operation log module, a data quality analysis module, a data processing module and a parameter calibration module;
[0061] The operation log module generates the position label and angle label of the focal plane based on the internal structure design of the infrared thermal imager, and records the control behavior information of the user end;
[0062] Wherein, the operation log module includes a cache library unit and a tag pair unit;
[0063] The cache unit is used to establish an operation log cache, which stores control behavior information for each dynamic adjustment of the imaging effect, and the control behavior information includes the position information and angle information of the focal plane; based on the internal structure design of the infrared thermal imager, a unified form of the position label and angle label of the focal plane are respectively established, and the position label and the angle label have unique label attributes respectively;
[0064] The label pair unit is used to record the i-th position label as , the jth angle label is recorded as , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as ,and ,in, represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, and E represents the total amount of control behavior information when the imaging effect is dynamically adjusted for the kth time, The position label indicating the focal plane and focal plane angle labels , the label pair formed;
[0065] The data quality analysis module is used to randomly divide the control behavior information into multiple groups, each of which constitutes a subset data cluster, and analyze the data quality to perform quality screening on the control behavior information;
[0066] Wherein, the data quality analysis module includes a subset composition unit and a quality analysis unit;
[0067] The subset constituent unit constructs a behavior representation two-dimensional coordinate system based on the control behavior information, wherein the horizontal coordinate of the behavior representation two-dimensional coordinate system corresponds to the position label of the focal plane, and the vertical coordinate of the behavior representation two-dimensional coordinate system corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, the control behavior information is punctuated in the behavior representation two-dimensional coordinate system; the control behavior set The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as ,and , , x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information As a subset of data clusters The centroids of the subset data clusters are obtained respectively The punctuation and control behavior information corresponding to each control behavior information The Euclidean distance between the corresponding punctuation points, based on the Euclidean distance, the subset data cluster is obtained The intra-cluster distance is denoted as ;
[0068] The quality analysis unit is used to build an iterative data quality analysis model. Randomly select a sample input, and the sample input selected in each iteration is different;
[0069] Controlling behavioral information As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows:
[0070]
[0071] in, Indicates control behavior information The data quality value of represents the expected value of the minimum intra-cluster distance;
[0072] Preset data quality threshold. If the data quality value is greater than or equal to the data quality threshold, the control behavior information is extracted. And save it in the output information set Otherwise, the control behavior information is removed ;
[0073] Until the control behavior set After every control behavior information in participates in the iteration, the iteration stops;
[0074] A data processing module classifies the control behavior information based on the screening results;
[0075] Wherein, the data processing module includes a first classifier unit and a second classifier unit;
[0076] The first classifier unit is used to output the information set Perform the first behavior classification, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, recorded as ,in, Represents the position information of the focal plane is the first classification reference, corresponding to the first classification set generated;
[0077] The second classifier unit is used to output the information set Perform the second behavior classification, use the angle information of the same focal plane as the second classification reference, traverse the label pairs with the same second classification reference, and generate the second classification set, recorded as ,in, Represents the angle information of the focal plane is the second classification reference, corresponding to the generated second classification set;
[0078] The parameter calibration module is used to evaluate the control behavior information through evaluation indicators and calibrate the control behavior based on the evaluation results.
[0079] Among them, the parameter calibration module includes an indicator parameter unit and a calibration analysis unit;
[0080] The indicator parameter unit is used to establish an evaluation indicator parameter category information library, which stores the indicator parameter category information for evaluating the regulation effect; the evaluation indicator parameters are uniformly coded to form an evaluation indicator parameter set with unique coding attributes, which is recorded as ,in, represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters;
[0081] Calibration analysis unit, used to evaluate the parameter set of indicators Each evaluation index parameter controls the behavior information Conduct assessments to obtain control behavior information Based on the initial evaluation mean of each control behavior information in the first classification set, the position information is obtained The comprehensive evaluation mean is recorded as ; Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information is obtained The comprehensive evaluation mean is denoted as ;
[0082] Respectively and The corresponding position labels and angle labels are used to calibrate the position and angle of the focal plane.
[0083] See also Figure 2 In the second embodiment, a multi-parameter abnormal calibration method for an infrared thermal imager is provided, the method comprising the following steps:
[0084] Step S100: Based on the internal structure design of the infrared thermal imager, generate a position label and an angle label of the focal plane, and record the control behavior information of the user end;
[0085] Specifically, an operation log cache is established, and the operation log cache stores control behavior information each time the imaging effect is dynamically adjusted, and the control behavior information includes the position information and the angle information of the focal plane; based on the internal structure design of the infrared thermal imager, a unified form of the focal plane position label and angle label are respectively established, and the position label and the angle label have unique label attributes respectively;
[0086] The i-th position label is recorded as , the jth angle label is recorded as , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as ,and ,in, represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, and E represents the total amount of control behavior information when the imaging effect is dynamically adjusted for the kth time, The position label indicating the focal plane and focal plane angle labels , the label pair formed;
[0087] Step S200: randomizing the control behavior information into multiple groups, each group forming a subset data cluster, and analyzing the data quality to perform quality screening on the control behavior information;
[0088] Specifically, based on the control behavior information, a behavior representation two-dimensional coordinate system is constructed, the horizontal coordinate of the behavior representation two-dimensional coordinate system corresponds to the position label of the focal plane, and the vertical coordinate of the behavior representation two-dimensional coordinate system corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, the control behavior information is punctuated in the behavior representation two-dimensional coordinate system; the control behavior set The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as ,and , , x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information As a subset of data clusters The centroids of the subset data clusters are obtained respectively The punctuation and control behavior information corresponding to each control behavior information The Euclidean distance between the corresponding punctuation points, based on the Euclidean distance, the subset data cluster is obtained The intra-cluster distance is denoted as ;
[0089] Construct a data quality iterative analysis model. In each iteration, Randomly select a sample input, and the sample input selected in each iteration is different;
[0090] Controlling behavioral information As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows:
[0091]
[0092] in, Indicates control behavior information The data quality value of represents the expected value of the minimum intra-cluster distance;
[0093] Preset data quality threshold. If the data quality value is greater than or equal to the data quality threshold, the control behavior information is extracted. And save it in the output information set Otherwise, the control behavior information is removed ;
[0094] Until the control behavior set After every control behavior information in participates in the iteration, the iteration stops;
[0095] Step S300: Classifying the control behavior information based on the screening result;
[0096] Specifically, for the output information set To classify the behavior, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, recorded as ,in, Represents the position information of the focal plane is the first classification reference, corresponding to the first classification set generated;
[0097] Taking the angle information of the same focal plane as the second classification reference, traverse the label pairs with the same second classification reference and generate the second classification set, which is recorded as ,in, Represents the angle information of the focal plane is the second classification reference, corresponding to the generated second classification set;
[0098] Step S400: Evaluate the control behavior information through evaluation indicators, and calibrate the control behavior based on the evaluation results;
[0099] Specifically, an evaluation indicator parameter category information database is established, in which the indicator parameter category information for evaluating the regulation effect is stored; the evaluation indicator parameters are uniformly coded to form an evaluation indicator parameter set with a unique coding attribute, which is recorded as ,in, represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters;
[0100] For example, the indicator categories include clarity, depth of field, distortion and aberration; at the same time, the parameters corresponding to each indicator category have diverse characteristics;
[0101] By evaluating the parameter set of the indicator Each evaluation index parameter controls the behavior information Conduct assessments to obtain control behavior information Based on the initial evaluation mean of each control behavior information in the first classification set, the position information is obtained The comprehensive evaluation mean is recorded as ; Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information is obtained The comprehensive evaluation mean is denoted as ;
[0102] Respectively and The corresponding position labels and angle labels are used to calibrate the position and angle of the focal plane;
[0103] For example, when the imaging effect is adjusted dynamically for the fifth time, a large number of control behaviors are regulated to obtain imaging effects under E=20 control behaviors. The algorithm performs adaptive learning based on these 20 effects and finally obtains and Corresponding location tag and angle labels , calibrate the position and angle of the focal plane.
[0104] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0105] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A multi-parameter abnormal calibration method for infrared thermal imagers, characterized in that: The method comprises the following steps: Step S100: Based on the internal structure design of the infrared thermal imager, generate a position label and an angle label of the focal plane, and record the control behavior information of the user end; Step S200: randomizing the control behavior information into multiple groups, each group forming a subset data cluster, and analyzing the data quality to perform quality screening on the control behavior information; Step S300: Classifying the control behavior information based on the screening result; Step S400: Evaluate the control behavior information through evaluation indicators, and calibrate the control behavior based on the evaluation results; The specific implementation process of step S100 includes: Step S101: Establishing an operation log cache library, wherein the operation log cache library stores control behavior information each time the imaging effect is dynamically adjusted, wherein the control behavior information includes the position information of the focal plane and the angle information of the focal plane; based on the internal structure design of the infrared thermal imager, establishing a unified form of the position label and the angle label of the focal plane, wherein the position label and the angle label have unique label attributes respectively; Step S102: record the i-th position label as S i , and record the jth angle label as U j , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as CB k ={tt e |e∈[1,E]}, and tt e =[S i , U j ], where tt e represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, E represents the total number of control behavior information when the imaging effect is dynamically adjusted for the kth time, [S i , U j ] indicates the position label S of the focal plane i and the focal plane angle label U j , the label pair formed; The specific implementation process of step S200 includes: Step S201: Based on the control behavior information, construct a behavior representation two-dimensional coordinate system, the horizontal coordinate of which corresponds to the position label of the focal plane, and the vertical coordinate of which corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, punctuate the control behavior information in the behavior representation two-dimensional coordinate system; for the control behavior set CB k The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as cb x ,and x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information tt e As a subset data cluster cb x The centroid of each control behavior information in the subset data cluster cbx is obtained, and the punctuation and control behavior information tt corresponding to each control behavior information are obtained. e The Euclidean distance between the corresponding punctuation points is used to obtain the subset data cluster cb based on the Euclidean distance. x The intra-cluster distance is denoted as D(cb x |tt e ); Step S202: Construct a data quality iterative analysis model. In each iteration, the control behavior set CB k Randomly select a sample input, and the sample input selected in each iteration is different; Control behavior information tt e As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows: Among them, DQ(tt e ) represents control behavior information tt e The data quality value of , dq represents the expected value of the minimum intra-cluster distance; Preset data quality threshold, if the data quality value is greater than or equal to the data quality threshold, then extract the control behavior information tt e And store it in the output information set OUT k Otherwise, the control behavior information tt is removed e ; Until the control behavior set CB k After every control behavior information in participates in the iteration, the iteration stops; The specific implementation process of step S300 includes: Step S301: Output information set OUT k To classify the behavior, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, denoted as FTS (S i ), where FTS(S i ) represents the position information S of the focal plane i is the first classification reference, corresponding to the first classification set generated; Step S302: Using the angle information of the same focal plane as the second classification reference, traverse the label pairs with the same second classification reference, and generate a second classification set, denoted as STS (U j ), where STS(U j ) represents the angle information U of the focal plane j is the second classification reference, corresponding to the generated second classification set; The specific implementation process of step S400 includes: Step S401: Establish an evaluation indicator parameter category information library, in which the evaluation indicator parameter category information for evaluating the regulation effect is stored; uniformly encode the evaluation indicator parameters to form an evaluation indicator parameter set with a unique encoding attribute, recorded as IC = {ic a |a∈[1,A]}, where ic a represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters; Step S402: Using each evaluation index parameter in the evaluation index parameter set IC to evaluate the control behavior information tt e Conduct evaluation and obtain control behavior information e Based on the initial evaluation mean of each control behavior information in the first classification set, the position information S is obtained. i The comprehensive evaluation mean is recorded as P1(S i ); Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information U is obtained j The comprehensive evaluation mean is denoted as P2(U j ); max{P1(S i )} and max{P2(U j )} The position and angle of the focal plane are calibrated by the corresponding position label and angle label.
2. A multi-parameter abnormal calibration system for infrared thermal imagers, characterized in that: The system includes: an operation log module, a data quality analysis module, a data processing module and a parameter calibration module; The operation log module generates a position label and an angle label of the focal plane based on the internal structure design of the infrared thermal imager, and records the control behavior information of the user end; The data quality analysis module is used to randomly divide the control behavior information into multiple groups, each group forming a subset data cluster, and analyze the data quality to perform quality screening on the control behavior information; The data processing module classifies the control behavior information based on the screening result; The parameter calibration module is used to evaluate the control behavior information through the evaluation index, and calibrate the control behavior based on the evaluation result; The operation log module includes a cache unit and a tag pair unit; The cache unit is used to establish an operation log cache, which stores control behavior information each time the imaging effect is dynamically adjusted, and the control behavior information includes the position information and the angle information of the focal plane; based on the internal structure design of the infrared thermal imager, a unified form of the focal plane position label and angle label are respectively established, and the position label and the angle label respectively have unique label attributes; The label pair unit is used to record the i-th position label as S i , and record the jth angle label as U j , then collect the control behavior information when dynamically adjusting the imaging effect for the kth time, and generate a control behavior set, recorded as CB k ={tt e |e∈[1,E]}, and tt e =[S i , U j ], where tt e represents the e-th control behavior information when the imaging effect is dynamically adjusted for the kth time, E represents the total number of control behavior information when the imaging effect is dynamically adjusted for the kth time, [S i , U j ] indicates the position label S of the focal plane i and the focal plane angle label U j , the label pair formed; The data quality analysis module includes a subset construction unit and a quality analysis unit; The subset forming unit constructs a behavior representation two-dimensional coordinate system based on the control behavior information, wherein the horizontal coordinate of the behavior representation two-dimensional coordinate system corresponds to the position label of the focal plane, and the vertical coordinate of the behavior representation two-dimensional coordinate system corresponds to the angle label of the focal plane; based on the behavior representation two-dimensional coordinate system, the control behavior information is punctuated in the behavior representation two-dimensional coordinate system; and the control behavior set CB is k The control behavior information in each control behavior is randomly grouped to form a subset data cluster, and any subset data cluster is obtained, which is recorded as cb x ,and x represents the number of the subset data cluster, y represents the total number of subset data clusters; to control the behavior information tt e As a subset data cluster cb x The centroid of each sub-set is obtained. x The punctuation and control behavior information corresponding to each control behavior information in tt e The Euclidean distance between the corresponding punctuation points is used to obtain the subset data cluster cb based on the Euclidean distance. x The intra-cluster distance is denoted as D(cb x |tt e ); The quality analysis unit is used to construct a data quality iterative analysis model. In each iteration, the control behavior set CB k Randomly select a sample input, and the sample input selected in each iteration is different; Control behavior information tt e As the sample input for the rth iteration, the data quality value is calculated. The specific calculation formula is as follows: Among them, DQ(tt e ) represents control behavior information tt e The data quality value of , dq represents the expected value of the minimum intra-cluster distance; Preset data quality threshold, if the data quality value is greater than or equal to the data quality threshold, then extract the control behavior information tt e And store it in the output information set OUT k Otherwise, the control behavior information tt is removed e ; Until the control behavior set CB k After every control behavior information in participates in the iteration, the iteration stops; The data processing module includes a first classifier unit and a second classifier unit; The first classifier unit is used to output the information set OUT k Perform the first classification, take the position information of the same focal plane as the first classification reference, traverse the label pairs with the same first classification reference, and generate the first classification set, recorded as FTS (S i ), where FTS(S i ) represents the position information S of the focal plane i is the first classification reference, corresponding to the first classification set generated; The second classifier unit is used to output the information set OUT k The second behavior is classified, and the angle information of the same focal plane is used as the second classification reference. The label pairs with the same second classification reference are traversed, and the second classification set is generated, which is recorded as STS (U j ), where STS(U j ) represents the angle information U of the focal plane j is the second classification reference, corresponding to the generated second classification set; The parameter calibration module includes an index parameter unit and a calibration analysis unit; The indicator parameter unit is used to establish an evaluation indicator parameter category information library, in which the evaluation indicator parameter category information for evaluating the regulation effect is stored; the evaluation indicator parameters are uniformly encoded to form an evaluation indicator parameter set with a unique encoding attribute, which is recorded as IC = {ic a |a∈[1,A]}, where ic a represents the ath evaluation index parameter, and A represents the total number of evaluation index parameters; The calibration analysis unit is used to evaluate the control behavior information tt by each evaluation index parameter in the evaluation index parameter set IC. e Conduct evaluation and obtain control behavior information e Based on the initial evaluation mean of each control behavior information in the first classification set, the position information S is obtained. i The comprehensive evaluation mean is recorded as P1(S i ); Based on the initial evaluation mean of each control behavior information in the second classification set, the angle information U is obtained j The comprehensive evaluation mean is denoted as P2(U j ); max{P1(S i )} and max{P2(U j )} The position and angle of the focal plane are calibrated by the corresponding position label and angle label.
Citation Information
Patent Citations
Imaging system environment adaptability evaluation method based on infrared simulation
CN116305705A
Label quality intelligent analysis method and device, electronic equipment and storage medium
CN116720118A
Infrared therapeutic instrument intelligent control method and system
CN117323578A
Infrared therapeutic instrument condition evaluation and residual life prediction method and system
CN117563144A
Tag domain presentation device, tag domain presentation method, and information processing system using the same
US20230097665A1
