Multimodal target detection data fusion method, electronic device and storage medium

Abstract

The present invention provides a multimodal target detection data fusion method, electronic device and storage medium, which relates to the technical field of multimodal target detection data fusion. The method comprises: obtaining the detection results of each preset detection model on the target area to obtain a detection result list M; determining the support of all preset detection models for the i-th preset detection model; determining the confidence Cred (M of the i-th preset detection model) of the i-th preset detection model. i ); get uncertainty list E; for M i Correction is performed to obtain a corrected detection result list M'; M' is fused with M to obtain a fusion result corresponding to the detection result of each preset detection model for the target area; the present invention can fuse multimodal target detection data into a final result, avoiding the situation where the target detection result is erroneous due to an abnormality of a preset detection result, thereby making the detection result of the target in the target area more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of multimodal target detection data fusion, and in particular to a multimodal target detection data fusion method, electronic equipment, and storage medium. Background Art

[0002] When identifying targets in a target area, in order to improve the accuracy of recognition, multiple types of recognition models are usually used to identify targets in the same area. The recognition results output by each type of recognition model are then fused, and the fusion results are used to comprehensively judge whether there is a target in the target area. Therefore, the accuracy of the fusion result will directly affect the judgment result of the target in the target area. How to accurately fuse multimodal target detection data has become an urgent problem to be solved. Summary of the Invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is:

[0004] According to a first aspect of the present application, a multimodal object detection data fusion method is provided, the method comprising the following steps:

[0005] S100, obtaining the detection results of each preset detection model on the target area to obtain a detection result list M=(M1, M2, ..., M i ,…M n ), i = 1, 2, ..., n; where M i is the detection result of the target area by the i-th preset detection model, n is the number of preset detection models; M i =(m i，1 , m i，2 );m i，1 is the probability that the target object exists in the target area detected by the i-th preset detection model, m i，2 The probability that the target object does not exist in the target area detected by the i-th preset detection model.

[0006] S200, based on M, determine the support Sup(M) of all preset detection models to the i-th preset detection model i ); Among them, Sup(M i ) is obtained based on the 1-norm distance between two preset detection models in M.

[0007] S300, according to Sup(M i ), determine the confidence Cred(M i )=Sup(M i ) / ∑ n i＝ 1Sup(M i ).

[0008] S400, based on M, determine the uncertainty of the detection result corresponding to each preset detection model to obtain an uncertainty list E = (E1, E2, ..., E i ,…,E n ); where E i is the uncertainty of the detection result corresponding to the i-th preset detection model; E i =-(m i，1 ×log2m i，1 +m i，2 ×log2m i，2 ).

[0009] S500, according to E and Cred(M i ), for M i Correction is performed to obtain the corrected detection result list M'=(M'1, M'2, ..., M' i ,…M' n ); where M' i M i The corresponding corrected detection results.

[0010] S600: Using the fusion rule corresponding to the DS evidence theory, M' is fused with M to obtain a fusion result corresponding to the detection result of each preset detection model on the target area.

[0011] According to another aspect of the present application, a non-transitory computer-readable storage medium is also provided, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement the above-mentioned multimodal target detection data fusion method.

[0012] According to another aspect of the present application, an electronic device is provided, including a processor and the above-mentioned non-transitory computer-readable storage medium.

[0013] The present invention has at least the following beneficial effects:

[0014] The multimodal target detection data fusion method of the present invention is improved based on the DS evidence theory to obtain the detection results corresponding to each preset detection model, that is, the original probability distribution; first, based on the original probability distribution, the confidence Cred (M i ); secondly, according to M, determine the uncertainty of the detection results corresponding to each preset detection model to obtain an uncertainty list E; according to E and Cred(M i ), for M iThe correction is performed to obtain a corrected detection result list M'; finally, the fusion rule corresponding to the DS evidence theory is used to fuse M' with M to obtain a fusion result corresponding to the detection result of each preset detection model for the target area; in the present invention, through the above method, the multimodal target object detection data can be fused into a final result, avoiding the situation where the target object detection result is erroneous due to an abnormality of a preset detection result, thereby making the detection result of the target object in the target area more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 Flowchart of a multimodal object detection data fusion method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0018] It should be noted that, based on this disclosure, those skilled in the art will appreciate that an aspect described herein can be implemented independently of any other aspect, and that two or more of these aspects can be combined in various ways. For example, any number of the aspects described herein can be used to implement an apparatus and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement such an apparatus and / or practice such a method.

[0019] The following will refer to Figure 1 The flowchart of the multimodal target detection data fusion method shown in FIG. 1 introduces a multimodal target detection data fusion method.

[0020] The multimodal object detection data fusion method may include the following steps:

[0021] S100, obtaining the detection results of each preset detection model on the target area to obtain a detection result list M=(M1, M2, ..., M i ,…M n), i = 1, 2, ..., n; where M i is the detection result of the target area by the i-th preset detection model, n is the number of preset detection models; M i =(m i，1 , m i，2 );m i，1 is the probability that the target object exists in the target area detected by the i-th preset detection model, m i，2 The probability that the target object does not exist in the target area detected by the i-th preset detection model.

[0022] In this embodiment, the preset detection models include a medium-wave infrared detection model, a long-wave infrared detection model, a radar wave detection model, and a visible light detection model; each type of detection model can output the probability of the target existing and the probability of the target not existing when detecting whether there is a target in the target area; m p，1 +m p，2 =1.

[0023] It should be noted that each preset detection model is equivalent to the evidence body in DS evidence theory, and the probability of the target object existing and the probability of the target object not existing in the detection results are equivalent to the support of the evidence body for the proposition in the recognition framework; thus, the corresponding original basic probability distribution can be obtained; for example, the original basic probability is shown in Table 1:

[0024] Table 1

[0025]

[0026] In Table 1, 0.81 represents the probability that the medium-wave infrared detection model of the evidence source detects that the target object exists in the target area, which can be understood as the support degree of the medium-wave infrared detection model of the evidence body for the proposition that the target object exists.

[0027] S200, based on M, determine the support Sup(M) of all preset detection models to the i-th preset detection model i ); Among them, Sup(M i ) is obtained based on the 1-norm distance between two preset detection models in M.

[0028] It should be noted that the 1-norm can effectively reduce the impact of conflict and noise on the preset detection model. Therefore, in this application, the 1-norm is used to calculate the distance conflict measurement between different preset detection models, that is, evidence bodies.

[0029] Specifically, step S200 may include the following steps:

[0030] S210, determining the 1-norm distance d between the i-th preset detection model and the j-th preset detection model according to M i，j =|mi，1 -m j，1 |+|m i，2 -m j，2 |; where j = 1, 2,…, n.

[0031] In this embodiment, the 1-norm distance d between the i-th preset detection model and the j-th preset detection model i，j It is obtained according to the existing 1-norm distance calculation formula and will not be described here.

[0032] S220, for d i，j Normalize to get d i，j The corresponding normalized 1-norm distance d' i，j =d i，j / 4+0.5.

[0033] In this embodiment, the probability of the target object existing and the probability of the target object not existing included in the detection results of any preset detection model are both within the range of 0-1; therefore, d i，j The value range is 0-2; based on this, d i，j Normalize to [0.5, 1] ​​for subsequent processing.

[0034] S230, according to d' i，j , determine Sup(M i )=2×∑ n j＝1 (d' i，j ×log2(1 / d' i，j )).

[0035] In this embodiment, through the above steps, the support degree Sup(M) of all preset detection models to the i-th preset detection model can be determined. i ).

[0036] S300, according to Sup(M i ), determine the confidence Cred(M i )=Sup(M i ) / ∑ n i＝ 1Sup(M i ).

[0037] In this embodiment, the confidence Cred(M i ) is the Sup(M i ) accounts for each preset detection model corresponding to Sup(M i ), Cred(M i) is larger, indicating that the detection result output by the i-th preset detection model is more credible; at this point, the calculation of the first branch in this embodiment is completed.

[0038] Then, the second branch is calculated through fuzzy preference relations. Fuzzy preference relations are a framework for handling decision problems. The basic probability allocation (BPA) is modified based on fuzzy preference relations of various information entropies. First, the Deng entropy values ​​of each preset detection model, that is, the evidence body, are calculated. The uncertainty of the basic probability allocation is calculated through the Deng entropy values, as follows:

[0039] S400, based on M, determine the uncertainty of the detection result corresponding to each preset detection model to obtain an uncertainty list E = (E1, E2, ..., E i ,…,E n ); where E i is the uncertainty of the detection result corresponding to the i-th preset detection model; E i =-(m i，1 ×log2m i，1 +m i，2 ×log2m i，2 ).

[0040] In this embodiment, the uncertainty of the detection result corresponding to each preset detection model is calculated based on the calculation formula of the Deng entropy value, which is not repeated here; after obtaining E, the variance of the Deng entropy value is calculated to construct the preference matrix.

[0041] S500, according to E and Cred(M i ), for M i Correction is performed to obtain the corrected detection result list M'=(M'1, M'2, ..., M' i ,…M' n ); where M' i M i The corresponding corrected detection results.

[0042] Specifically, step S500 may include the following steps:

[0043] S510: According to E, determine the intermediate value corresponding to each uncertainty in E to obtain an intermediate value list V = (V1, V2, ..., V i ,…,V n ); where V i For E i The corresponding median value; e is a natural constant.

[0044] S520, according to V, determine the variance corresponding to each uncertainty in E to obtain a variance list VA = (VA1, VA2, ..., VAi ,…,VA n ); among them, VA i For E i Corresponding variance; VA i =Var(V1, V2, ..., V i-1 , V i+1 ,…,V n ); Var() is the preset variance function.

[0045] It should be noted that when calculating VA i When V i Those skilled in the art can use existing variance calculation methods to determine the variance corresponding to each uncertainty in E according to actual needs, which will not be elaborated here.

[0046] S530, constructing a corresponding fuzzy preference matrix P according to VA; wherein P includes n rows and 2 columns; the element ρ in the i-th row and k-th column of P i，k =VA i / (VA i +VA k ); k=1,2.

[0047] In this embodiment, the recognition framework includes two propositions, namely, the existence of the target object and the absence of the target object. Therefore, P includes two columns.

[0048] S540, based on P, determine the consistency matrix P' corresponding to P; wherein P' includes n rows and 2 columns; the element ρ' in the i-th row and k-th column of P' i，k =(1 / n)×∑ 2 r＝1 (ρ i，k +ρ k，r ); r=1,2.

[0049] S550: Determine the consistency degree HK corresponding to P' according to P'.

[0050] Furthermore, step S550 may include the following steps:

[0051] S551, according to P', determine E i The corresponding average consistency μ i =(1 / 2)×∑ 2 k＝1 ρ' i，k .

[0052] S552, according to μ i , determine the maximum average consistency α=MAX(μ i |1≤i≤n) and minimum average consistency β=MIN(μ i|1≤i≤n); wherein, MAX() is a preset maximum value function, and MIN() is a preset minimum value function.

[0053] S553, based on α and β, determine the corresponding range constant γ = 1 / (2×MAX(0.5, (α-β))).

[0054] In this embodiment, γ is a range constant, so that the preference value of the fuzzy preference consistency matrix is ​​between (0, 1].

[0055] S554, based on P and P', determine the consistency between P and P' λ = 1-(2 / (n×(n-1)))∑ n i＝1 ∑ 2 k＝1，k≠i |ρ i，k -ρ' i，k |.

[0056] In this embodiment, λ represents the degree of consistency between the fuzzy preference matrix and the fuzzy preference consistency matrix. The larger λ is, the more consistent the fuzzy preference matrix and the fuzzy preference consistency matrix are.

[0057] S555, based on γ and λ, determine HK = γ × λ.

[0058] S560, according to HK, modify the elements in P' to obtain a modified consistency matrix XP' corresponding to P'; wherein XP' includes n rows and 2 columns; the element Qρ' in the i-th row and k-th column of XP' i，k =ρ' i，k ×HK+(1-HK) / 2.

[0059] S570, according to XP', determine E i The corresponding ranking value R i =(2 / (n 2 -n))×∑ 2 k＝1，k≠i Qρ' i，k .

[0060] S580, according to Cred(M i ) and R i , for M i Make corrections to get M' i .

[0061] Furthermore, step S580 may include the following steps:

[0062] S581, according to Cred(M i ) and R i , determine the correction weight W corresponding to the i-th preset detection model joint (Mi )=Cred(M i )×R i / ∑ n i＝1 (Cred(M i )×R i ).

[0063] S582, according to W joint (M i ), for M i Make corrections to get M' i =(m' i，1 , m' i，2 ); where m' i，1 is m i，1 The corresponding modified probability, m' i，2 is m i，2 The corresponding modified probability; m' i，1 =m i，1 ×W joint (M i );m' i，2 =m i，2 ×W joint (M i ).

[0064] In this embodiment, the original basic probability distribution is corrected through the method in the above steps. The correction process combines the detection results of each preset detection model, so that the subsequent fusion result can be more accurate.

[0065] S600: Using the fusion rule corresponding to the DS evidence theory, M' is fused with M to obtain a fusion result corresponding to the detection result of each preset detection model on the target area.

[0066] Furthermore, step S600 may include the following steps:

[0067] S610, traverse M', if m' i，1 =0, then set m' i，1 =0.001, set m' i，2 =0.999; if m' i，2 =0, then set m' i，2 =0.001, set m' i，1 =0.999; to obtain the target detection result list WM corresponding to M.

[0068] S620: Use the fusion rule corresponding to the DS evidence theory to fuse M and WM.

[0069] S630, if m' i，1 ≠0 and m' i，2≠0, then use the fusion rule corresponding to the DS evidence theory to fuse M' with M.

[0070] In this embodiment, the fusion rule of DS evidence theory can be used to fuse the revised basic probability distribution, i.e., M', with the original basic probability distribution, i.e., M, to obtain a fusion result. Prior to this, since the fusion rule is still the fusion rule corresponding to the existing DS evidence theory, even if the original probability distribution is revised, if there is 0 in the probability, a veto phenomenon will still exist. In order to avoid the veto phenomenon, the above step S610 is executed, thereby effectively avoiding the veto phenomenon.

[0071] The explanation for the veto is as follows:

[0072] Table 2 shows examples of conflicting evidence.

[0073] Table 2

[0074]

[0075] As shown in Table 2, for the high-conflict and complete-conflict examples, in the high-conflict example, the conflict factor TK between the evidence bodies calculated using the existing fusion rules of DS evidence theory is 0.999, and the final fusion result is: m1⊕m2(A)=0, m1⊕m2(B)=1, m1⊕m2(C)=0. The results show that the two evidence bodies m1 and m2 have the lowest support for Proposition B, but the final fusion result is an absurd result, that is, it fully supports Proposition B. In the example of complete conflict, the conflict factor TK=1, and the final fusion result is: m1⊕m2(A)=0, m1⊕m2(B)=0, m1⊕m2(C)=0. From the final result, it can be seen that the fusion result denies the three propositions A, B and C. At this time, the existing DS evidence theory has become invalid, and a veto phenomenon has occurred, that is, if the basic probability distribution BPA of any evidence body for a certain proposition is 0, then the BPA of the proposition after fusion is also 0.

[0076] The multimodal target detection data fusion method of this embodiment is improved based on the DS evidence theory to obtain the detection results corresponding to each preset detection model, that is, the original probability distribution; first, based on the original probability distribution, the confidence Cred (M i ); secondly, according to M, determine the uncertainty of the detection results corresponding to each preset detection model to obtain an uncertainty list E; according to E and Cred(M i ), for M iThe correction is performed to obtain a corrected detection result list M'; finally, the fusion rule corresponding to the DS evidence theory is used to fuse M' with M to obtain a fusion result corresponding to the detection result of each preset detection model for the target area; in the present invention, through the above method, the multimodal target object detection data can be fused into a final result, avoiding the situation where the target object detection result is erroneous due to an abnormality of a preset detection result, thereby making the detection result of the target object in the target area more accurate.

[0077] Furthermore, the probability of 0 in the corrected probability is corrected again to avoid the veto phenomenon and further improve the accuracy and rationality of multimodal target detection data fusion.

[0078] Furthermore, although the steps of the method of the present disclosure are described in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in this particular order, or that all steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0079] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0080] The program product may be implemented in any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0081] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0082] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0083] The program code for performing the operations of the present application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0084] An embodiment of the present invention further provides an electronic device including a processor and the aforementioned non-transitory computer-readable storage medium.

[0085] The electronic device is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0086] The electronic device is implemented as a general-purpose computing device. Components of the electronic device may include, but are not limited to, the aforementioned at least one processor, the aforementioned at least one memory, and a bus connecting different system components (including the memory and the processor).

[0087] The memory stores program codes, which can be executed by the processor, so that the processor performs the steps of various embodiments described in this specification.

[0088] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).

[0089] The memory may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0090] The bus may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

[0091] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface. Furthermore, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the electronic device via a bus. It should be understood that, although not shown in the figures, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0092] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.

[0093] An embodiment of the present invention further provides a computer program product comprising program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.

[0094] Although some specific embodiments of the present invention have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A multimodal target detection data fusion method, characterized in that: The method comprises the following steps: S100, obtaining the detection results of each preset detection model on the target area to obtain a detection result list M=(M1, M2, ..., M i ,…M n ), i=1, 2, ..., n; where M i is the detection result of the target area by the i-th preset detection model, n is the number of preset detection models; M i =(m i，1 , m i，2 );m i，1 is the probability that the target object exists in the target area detected by the i-th preset detection model, m i，2 is the probability that the target object does not exist in the target area detected by the i-th preset detection model; different preset detection models use different modalities, and each preset detection model targets the same target object; the preset detection models include medium-wave infrared detection model, long-wave infrared detection model, radar wave detection model and visible light detection model; m i，1 +m i，2 =1; S200, based on M, determine the support Sup (M) of all preset detection models to the i-th preset detection model i ); Among them, Sup (M i ) is obtained based on the 1-norm distance between two preset detection models in M; S300, according to Sup (M i ), determine the confidence Cred (M i ) = Sup (M i ) / ∑ n i=1 Sup(M i ); S400, based on M, determine the uncertainty of the detection result corresponding to each preset detection model to obtain an uncertainty list E = (E1, E2, ..., E i ,…,E n ); where E i is the uncertainty of the detection result corresponding to the i-th preset detection model; E i =-(m i，1 ×log2m i，1 +m i，2 ×log2m i，2 ); S500, according to E and Cred (M i ), for M i Make corrections to obtain the corrected detection result list M'=(M'1, M'2, ..., M' i ,…M' n ); where M' i M i The corresponding corrected test results; S600, using the fusion rule corresponding to the DS evidence theory, fusing M' and M to obtain a fusion result corresponding to the detection result of each preset detection model on the target area; Step S600 includes the following steps: S610, traverse M', if m' i，1 =0, then set m' i，1 =0.001, set m' i，2 =0.999; if m' i，2 =0, then set m' i，2 =0.001, set m' i，1 =0.999; to obtain the target detection result list WM corresponding to M; S620, using the fusion rule corresponding to the DS evidence theory, fuse M and WM; S630, if m' i，1 ≠0 and m' i，2 ≠0, then use the fusion rule corresponding to the DS evidence theory to fuse M' with M.

2. The multimodal target detection data fusion method according to claim 1, characterized in that: Step S200 includes the following steps: S210, determining the 1-norm distance d between the i-th preset detection model and the j-th preset detection model according to M i，j =|m i，1 -m j，1 |+|m i，2 -m j，2 |; where j = 1, 2, …, n; S220, for d i，j Normalize to get d i，j The corresponding normalized 1-norm distance d' i，j =d i，j / 4+0.5; S230, according to d' i，j , determine Sup (M i ) = 2 × ∑ n j=1 (d' i，j ×log2(1 / d' i，j )).

3. The multimodal target detection data fusion method according to claim 1, characterized in that: Step S500 includes the following steps: S510, according to E, determine the intermediate value corresponding to each uncertainty in E to obtain an intermediate value list V = (V1, V2, ..., V i ,…,V n ); where V i For E i The corresponding intermediate value; V i = ;e is a natural constant; S520, based on V, determine the variance corresponding to each uncertainty in E to obtain a variance list VA = (VA1, VA2, ..., VA i ,…,VA n ); Among them, VA i For E i Corresponding variance; VA i =Var(V1, V2, …, V i-1 , V i+1 ,…,V n ); Var() is the preset variance function; S530, constructing a corresponding fuzzy preference matrix P according to VA; wherein P includes n rows and 2 columns; the element ρ in the i-th row and k-th column of P i，k =VA i / (VA i +VA k );k=1,2; S540, based on P, determine the consistency matrix P' corresponding to P; wherein P' includes n rows and 2 columns; the element ρ' in the i-th row and k-th column of P' i，k =(1 / n)×∑ 2 r=1 (ρ i，k +ρ k，r );r=1,2; S550, based on P', determining the consistency degree HK corresponding to P'; S560, according to HK, modify the elements in P' to obtain a modified consistency matrix XP' corresponding to P'; wherein XP' includes n rows and 2 columns; the element Qρ' in the i-th row and k-th column of XP' i，k =ρ' i，k ×HK+(1-HK) / 2; S570, according to XP', determine E i The corresponding ranking value R i =(2 / (n 2 -n))×∑ 2 k=1，k≠iQ ρ' i，k ; S580, according to Cred (M i ) and R i , for M i Make corrections to get M' i .

4. The multimodal target detection data fusion method according to claim 3, characterized in that: HK is determined through the following steps: S551, according to P', determine E i The corresponding average consistency μ i =(1 / 2)×∑ 2 k=1 ρ' i，k ; S552, according to μ i , determine the maximum average consistency α=MAX(μ i |1≤i≤n) and minimum average consistency β=MIN(μ i |1≤i≤n); where MAX() is the preset maximum value function, and MIN() is the preset minimum value function; S553, based on α and β, determine the corresponding range constant γ = 1 / (2 × MAX(0.5, (α - β))); S554, based on P and P', determine the consistency between P and P' λ = 1-(2 / (n×(n-1)))∑ n i=1 ∑ 2 k=1，k≠i |ρ i，k -ρ' i，k |; S555, based on γ and λ, determine HK=γ×λ.

5. The multimodal target detection data fusion method according to claim 3, characterized in that: Step S580 includes the following steps: S581, according to Cred (M i ) and R i , determine the correction weight W corresponding to the i-th preset detection model joint (M i ) = Cred (M i ) × R i / ∑ n i=1 (Cred(M i ) × R i ); S582, according to W joint (M i ), for M i Make corrections to get M' i =(m' i，1 , m' i，2 ); where m' i，1 is m i，1 The corresponding modified probability, m' i，2 is m i，2 The corresponding modified probability; m' i，1 =m i，1 ×W joint (M i );m' i，2 =m i，2 ×W joint (M i ).

6. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the multimodal target detection data fusion method as described in any one of claims 1 to 5.

7. An electronic device, characterized in that: The device comprises a processor and the non-transitory computer-readable storage medium of claim 6.

Images