A two-dimensional and three-dimensional map adaptive projection matching method based on hidden Markov chain

Abstract

The application discloses a two-three-dimensional map adaptive projection matching method based on a hidden Markov chain, constructs a hidden Markov projection prediction model based on user preferences and projection characteristics, takes a map projection category as a hidden state, takes a user browsing position or a visual angle as an observation state, fuses user preferences and projection standard parameters to form a probability of each hidden state occurrence at each moment, updates an observation probability at each moment, and forms a current optimal projection matching combination through continuous iteration. The application integrates quantitative projection characteristics, combines user preferences, and influences a two-three-dimensional GIS visualization mode in a brand-new concept, selects a map projection suitable for different users and different scenes from an intelligent perspective, effectively overcomes browsing obstacles caused by a lack of projection field knowledge of part of users, and enhances a user's visualization experience.

Description

Technical Field

[0001] This invention patent belongs to the field of map visualization and relates to an adaptive projection matching method for two-dimensional and three-dimensional maps based on hidden Markov chains. Background Technology

[0002] The core objective of adaptive projection is to improve the accuracy of maps at global scales while ensuring a superior user experience. As is well known, all projections introduce distortions related to shape, size, and distance; therefore, choosing the appropriate projection is an art of trade-offs. To achieve intelligent projection matching, integrated 2D and 3D platforms need to fully consider user preferences and projection accuracy to form reasonable projection-assisted decisions. With the rapid development of cartography, various map projection methods with distinct characteristics have emerged. Different regions, scenarios, and users have significantly different projection requirements. For example, to display more accurate territorial boundaries, equal-area projection is needed, while common projection methods in Antarctica include polar azimuth stereoprojection, transverse Mercator projection, and Lambert conformal conic projection. To represent phenomena that change with longitude, the Paterson projection is considered. Traditional projection switching methods are mostly manual, which can meet a wide range of map projection needs, but not all users have a deep understanding of projection characteristics. Therefore, promoting research on intelligent projection matching aligns with the development concept of next-generation GIS.

[0003] In China, for seamless projection transformation, Niu Ruitao et al. established a projection library based on Proj.4. Regarding projection selection and matching, research has proposed a map projection method based on case-based analogical reasoning and a map projection selection method based on information hierarchy. The former uses the projection selection result as a reference standard and constructs a case-based database to analyze case application scenarios and recommend suitable map projections; the latter refines and quantifies the factors influencing map projection selection by clarifying the information hierarchy of the map projection selection problem, thus achieving computer-aided map projection selection. Internationally, PC Gosling et al. proposed an automatic map selection scheme combining projection distortion measurement and visual evaluation, which can improve users' understanding of the projection selection process. B et al. explored user preferences, providing additional criteria for selecting world map projections, effectively advancing the process of projection automation, and discussed how adaptive composite map projection schemes can incorporate multiple other equal-area projections. However, the above work did not consider user preferences for projections during viewpoint roaming, such as the choice of projection at a certain observation position and the projection fidelity requirements (isotropy, area, buckling, skewness, distance, and boundary cutting characteristics). Summary of the Invention

[0004] The purpose of this invention is to propose an adaptive projection matching method for two-dimensional and three-dimensional maps based on hidden Markov chains.

[0005] The technical solution to achieve the purpose of this invention is as follows: an adaptive projection matching method for two-dimensional and three-dimensional maps based on hidden Markov chains, comprising the following steps:

[0006] Step 101: Construct a Hidden Markov Projection Prediction Model based on user preferences and projection characteristics. The Hidden Markov Projection Prediction Model based on user preferences and projection characteristics takes the map projection category as the hidden state and the user's browsing position or viewpoint as the observed state. It integrates the user's preference for using projection and projection standardization parameters to form the probability of each hidden state occurring at each time. It updates the observation probability at each time and iterates continuously to form the current optimal projection matching combination.

[0007] Step 102: Initialize the time series {1, 2, ..., T} of a single user browsing session, and the initial observation sequence. Projection sequence set Map projection comprehensive evaluation error User preference prediction and projection evaluation characteristic probability ratio ξ, optimal matching projection sequence

[0008] Step 103: Integrate user preferences and projection characteristics to calculate the probability value π of the initial hidden state being ζ. ζ ;

[0009] Step 104: Calculate the maximum probability δ1(ζ) among all paths with the initial hidden state ζ;

[0010] Step 105: Calculate the observation probability b′ at the initial time. ζ (o1);

[0011] Step 106: Based on the principle of the Viterbi algorithm, iteratively calculate the maximum probability δ among all paths with state ζ at time t. t (ζ) and the previous state that determines the path with the highest probability are The probability ψ t (ζ), determine the Viterbi path-hidden state sequence that is most likely to produce the sequence of observed events;

[0012] Step 107: Calculate the optimal projection matching path

[0013] Furthermore, in step 103, by integrating user preferences and projection characteristics, the probability value π of the initial state being ζ is calculated. ζ The specific method is as follows:

[0014] Six error metrics are used: isotropy I, area A, curvature F, skewness S, distance D, and boundary cut B, with a standardized constant N. i =0.51, N a =0.41, N f =0.64, N s =0.60, N d =0.449, N b =0.25, then the comprehensive evaluation error S of the planar map projection is... e Represented as:

[0015]

[0016] For projection library Let the probability of each state occurring at the initial time be Π={π1,π2,...,π n}, where π i Indicates based on projection p i The initial state probability is defined as π, where π is the state probability determined by user preferences. i′ The state probability determined by the projection property is π. i ",and π i′ With the weight ξ assuming 0 ≤ ξ ≤ 1, we obtain:

[0017] π i =ξπ i′ +(1-ξ)π i″

[0018] Where, π i″ The error was obtained by normalizing the comprehensive evaluation error of the planar map projection. Let the projection sequence involved in a single map browsing be... Corresponding planar map projection comprehensive evaluation error The normalized comprehensive evaluation error of the planar map projection is expressed as Π(S)={π(S1), π(S2), ..., π(S2)}. k Since the comprehensive evaluation error value is inversely proportional to the similarity of the projection, the following holds:

[0019] S e1 π(S1)=S e2 π(S2)=...=S ek π(S k )

[0020] π(S1)+π(S2)+…+π(S k ) = 1

[0021] We can deduce that:

[0022]

[0023] Through π(S) i ) Calculate π i″ Therefore, we get:

[0024]

[0025] Therefore, the initial state is obtained as ζ = p. i The probability value π ζ .

[0026] Further, in step 104, the maximum probability δ1(ζ) among all paths with an initial hidden state ζ is calculated, specifically as follows:

[0027] δ1(ζ)=π ζ b ζ (o1)

[0028] Among them, b ζ (o1) represents the probability that the hidden state is ζ and the observed state is o1.

[0029] Further, in step 105, determine the previous hidden state of the path with the highest probability at the initial time. The probability ψ1(ζ) is calculated using the following method:

[0030] ψ1(ζ)=0.

[0031] Further, in step 105, the observation probability b′ at the initial time is calculated. ξ (o1), the specific method is as follows:

[0032]

[0033] Furthermore, in step 106, based on the principle of the Viterbi algorithm, the maximum probability δ among all paths with state ζ at time t is iteratively calculated. t (ζ) and the previous state of the path with the highest probability are The probability ψ t (ζ), determine the Viterbi path-hidden state sequence most likely to produce the sequence of observed events, specifically using the following method:

[0034]

[0035]

[0036] in, The hidden state is indicated by The probability of converting to ζ.

[0037] Further, step 107: Calculate the optimal projection matching path. The specific method is as follows:

[0038] Let P be the probability of finding the optimal path. * The optimal path destination is So

[0039]

[0040]

[0041] For t = T-1, T-2, ..., 1, optimal path backtracking:

[0042]

[0043] Find the optimal path

[0044] An adaptive projection matching system for 2D and 3D maps based on Hidden Markov Chains is provided, which implements the aforementioned adaptive projection matching method for 2D and 3D maps based on Hidden Markov Chains to achieve adaptive projection matching of 2D and 3D maps based on Hidden Markov Chains.

[0045] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned adaptive projection matching method for two-dimensional and three-dimensional maps based on hidden Markov chains, thereby achieving adaptive projection matching for two-dimensional and three-dimensional maps based on hidden Markov chains.

[0046] A computer-readable storage medium storing a computer program thereon, wherein when the computer program is executed by a processor, the aforementioned adaptive projection matching method for two-dimensional and three-dimensional maps based on hidden Markov chains is implemented to achieve adaptive projection matching of two-dimensional and three-dimensional maps based on hidden Markov chains.

[0047] Compared with existing technologies, the significant advantages of this invention are as follows: It adopts the concept of Hidden Markov Chain Programming to construct an adaptive map projection matching model based on a two-dimensional and three-dimensional integrated visualization. Due to the differences between map projections, this model incorporates quantitative projection characteristics and combines user preferences to influence the visualization method of two-dimensional and three-dimensional GIS with a brand-new concept. From an intelligent perspective, it selects map projections that are suitable for different users and different scenarios, effectively overcoming the browsing obstacles caused by some users' lack of knowledge in the field of projection, and enhancing the user's visualization experience. Attached Figure Description

[0048] Figure 1 It is a concept diagram of a two-dimensional and three-dimensional integrated visual adaptive map projection;

[0049] Figure 2 The mapping relationship between observed states and hidden states;

[0050] Figure 3 A schematic diagram of a hidden Markov chain stochastic process;

[0051] Figure 4 Therefore A schematic diagram of the ζ-path in the state space. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] Combination Figure 1 An adaptive projection matching method for 2D and 3D maps based on Hidden Markov Chains (HMMs) includes the following: establishing a projection library and creating unique indexes for potentially used map projections; collecting user preference samples for projection usage at different browsing locations; and constructing a projection prediction model (HMM) based on user preferences, with map projection categories as hidden states and user browsing locations or viewpoints as observed states. The mapping relationship between observed and hidden states is described in [the relevant documentation / section]. Figure 2 As shown in the figure. Based on six error measurement parameters of projection, including isotropy, area, curvature, skewness, distance, and boundary cutting, the similarity between the planar map projection and the sphere, i.e., map similarity, is comprehensively quantified to form important parameters for measuring the standardity of projection. The error values ​​of the six attributes of some standard projections are shown in Table 1, and the comprehensive evaluation quantification values ​​of some standard projections are shown in Table 2. By integrating user preferences for projection and projection standardity parameters through a preset adjustment factor, the probability of each state of the Hidden Markov Model at the initial time is formed through a normalization method. Similarly, the observation probability at each time is updated based on the adjustment factor; through continuous iteration, the current optimal projection matching combination is formed. The technical solution of this invention and the scientific principles on which it is based are described in detail below.

[0054] Table 1 lists the six major attribute errors of some standard projections.

[0055]

[0056]

[0057] Table 2: Quantitative Evaluation of Partial Standard Projections

[0058]

[0059] (1) Projective prediction model of hidden Markov chains based on user preferences (HMM)

[0060] The observation states of the Hidden Markov Model (HMM) based on user preferences are segments a, b, and c. A segment can be understood as an observation viewpoint in a different region at a different viewing height. Segment a is selected for projection. Section b projection selection Section C projection selection Wherein, projection p i Projection p j and projection p k In hidden state, For projection libraries.

[0061] Hidden random projection variables (HMMs) involve a set of hidden random projection variables. The observed segment sequence Q = {q1,q2,…,q} m}, section such Figure 2 a, b, and c in the implicit Markov chain. Each node in the implicit Markov chain. The prior results will produce an observable probability distribution on the observed variables, thus the stochastic processes of the nodes in the chain can be analyzed through the observed variables. Figure 3 This diagram illustrates a Hidden Markov Chain (HMM) based on user preference adaptive projection. Arrows in the diagram represent dependencies between variables. Considering user preferences for projection across different segments, the HMM relies on a strict assumption of observation independence, i.e., the observation result τ... i ∈Q depends only on the current state of the Markov chain.

[0062] The state of an Hidden Markov Model (HMM) at the next time step is determined solely by the current state and does not depend on any previous states. Based on this dependency, assuming the total number of states in the state space is k, the joint probability distribution of all variables is...

[0063]

[0064] In addition to structural information, the determination of the adaptive projective hidden Markov probability model also requires the following three sets of parameters:

[0065] State transition probability, that is, the probability of the model transitioning between different states, such as from projection. Transfer to section The state transition probability can be defined as matrix A = [a i，j ] n×n ,in

[0066] a i，j =P{ζ t+1 =p j |ζ t =p i}#(1)

[0067] a i，j Let p represent any time t, where the state is p. iThe state at the next moment is p j The probability is given by 1 ≤ i, j ≤ n.

[0068] Output observation probabilities: The Hidden Markov Model (HMM) obtains the probability of each observation based on the current state, defining the observation probability matrix B = [b i，j ] n×m ,in

[0069] b i，j =P(τ) t =q j |ζ t =p i )#(2)

[0070] b i，j This indicates that at any time t, the state is p. i Then the observed value q j The probability of being acquired, where i∈[1,n],j∈[1,m].

[0071] Initial state probability, the probability of each state occurring at the initial time of an Hidden Markov Model (HMM), is defined as Π = {π1, π2, ..., π}. n}, i∈[1, n], then

[0072] π i =P(ζ1=p i )#(3)

[0073] The purpose of adaptive projective hidden Markov probabilistic models is to find the state sequence that maximizes the probability of meeting given observation conditions. Assume the initial observation sequence of the HMM preference prediction model is Given observation sequence Seeking the optimal state sequence The conditional probability is

[0074] The state transition probability matrix A, the output observation probability matrix B, and the initial state probability sequence Π can all be obtained through sampling. Browser behavior related to projection transformations is collected; for example, when a user selects projection p. i p j and p k Perform a transformation browsing, assuming the projection transformation sequence is p. i →p j →p k →p i ,in The corresponding observation segment is q α →q β →q γ →q η , and {q α qβ q γ q η}∈Q, this process can form a record, and the projected state transition record p is statistically analyzed. i →p j →p k →p i As a sample of probability matrix A, the statistics p i p j and p k Status-based browsing segment scheme q α →q β →q γ →q η As a first-order sample of probability matrix B, statistical projection p i p j and p k The record selected as a sample in probability sequence Π is assumed to be in state p. i The total number of samples that caused the transition from the starting point is p i →p j State transition statistics are Then p i →p j The state transition probability is Similarly, suppose the HMM model is currently located at p i In projection mode, select q. α The number of times the section was browsed was And p i The total number of times the state occurs is Then the probability of observation Assuming projection sequence The total number of times the intermediate state occurs is So based on p i Initial state probability of a state

[0075] The problem of adaptive projection based on user preferences is actually about finding the best projection for each segment to meet the general needs of users. It is also the problem of optimal browsing path planning, i.e., the decoding problem of HMM.

[0076] To avoid local optima, the Viterbi algorithm is used to estimate the global optimum.

[0077] For ease of description, assume a given HMM model λ = (A, B, Π), a single time sequence of Earth exploration {1, 2, ..., T}, and a projected observation sequence. Define variable δ t (ζ) represents the maximum probability among all paths with state ζ at time t, where all paths contain elements defined as... Therefore, we can conclude the following:

[0078]

[0079] The path diagram for ζ is shown below. Figure 4 As shown.

[0080] Assume that at time t, the state variable ζ = p i State observation o t =q j Then it is easy to derive δ t (ζ)=π i b i，j ,in,

[0081] And from this, we can derive the recursive formula for δ as follows:

[0082]

[0083] b ζ (o t+1 () means the observed state is o t+1 The probability of the hidden state ζ; Represents the state transition probability, i.e., by The probability of converting to ζ.

[0084] For each state ζ∈Q, record the previous state of the path with the highest probability. The probability is

[0085]

[0086] ψ t (ζ) represents the conditional probability of being in state ζ at time t, which is the probability that the Hidden Markov Model is in state ζ given the observation sequence.

[0087] Assume t = 1, 2, ..., T, and define the probability of the optimal path as P. * The optimal path destination is So

[0088]

[0089]

[0090] For t = T-1, T-2, ..., 1, optimal path backtracking:

[0091]

[0092] The optimal path can be obtained.

[0093] Optimal path ζ * It can be used as a preference prediction for adaptive projection users to adjust their preferences under different observation states (browsing sections, see...) Figure 2 The projection (hidden state) is shown in the diagrams a, b, and c. The Hidden Model (HMM) can predict the optimal projection configuration given a browsing segment.

[0094] (2) Projection characteristics evaluation and quantification

[0095] To facilitate the evaluation of the similarity between a planar map projection and a sphere, Goldberg & Gott proposed six error metrics: isotropy (I), area (A), curvature (F), skewness (S), distance (D), and boundary cut (B), and provided the attribute errors of the standard projection, as shown in Table 1.

[0096] Peter Laskowski proposed a ranking weighting scheme based on map similarity, where the normalization parameter is the value in the rectangular projection, i.e., the normalization constant N. i =0.51, N a =0.41, N f =0.64, N s =0.60, N d =0.449, N b =0.25, then the overall evaluation error of the planar map projection is the square of the weighted parameter error, expressed as:

[0097]

[0098] (3) Normalization of multivariate properties of adaptive projection model

[0099] For projection sequences The probability of each state occurring at the initial time is Π={π1,π2,....,π n}, and based on p i Initial state probability of projection in For projection sequence The total number of times a state occurs. For ease of distinction, the probability of the initial state determined by sampled user behavior is defined as π. i′ The initial state probability determined by the projection properties is π. i″ , and π i′ With the presupposed proportion ξ, 0≤ξ≤1, it is easy to derive the proportion based on p i initial state probability

[0100] π i =ξπ i′ +(1-ξ)π i″ #(11)

[0101] The initial state probability π determined by the projection properties i″ This is obtained by normalizing the error parameters based on the comprehensive evaluation of projection similarity, that is, converting the comprehensive evaluation value of the projection into a decimal between (0, 1). Projection Library Assuming a single map browsing involves a sequence of projections... The corresponding planar map projection comprehensive evaluation error is Define the normalized initial state probability as Π(S) = {π(S1), π(S2), ..., π(S3)}. k Since the comprehensive evaluation error value is inversely proportional to the similarity of the projection, the following process exists:

[0102] S e1 π(S1)=S e2 π(S2)=...=S ek π(S k )=φ#(12)

[0103] π(S1)+π(S2)+...+π(S k )=1#(13)

[0104] Based on the derivation process (12) and (13), it is easy to conclude that

[0105]

[0106]

[0107] Equation (12) can be derived Where 1≤i≤k,

[0108]

[0109] Comprehensive evaluation error value sequence based on planar map projection The calculation results are shown in Table 2.

[0110] The normalized initial state probability value Π(S)={π(S1), π(S2), ..., π(S3)} can be calculated using formula (15). k The initial HMM state probability value that integrates user preferences and projection characteristics can be calculated from formula (11).

[0111]

[0112]

[0113]

[0114] The state transition probabilities of a multivariate MC-HMM (Hidden Markov Projective Prediction Model based on user preferences and projection characteristics) are completely determined by user preferences. The observation probabilities are calculated in the same way as the probabilities of each state occurring at the initial time.

[0115]

[0116] (4) Viterbi Algorithm Principle

[0117] Suppose we have a hidden Markov model (HMM) with a state space S containing k states, and the probability of the initial state i is π. i The transition probability from state i to state j is a. i，j Let the observed outputs be y1, ..., y T The most likely state sequence x1, ..., x2 that produces the observed result. T Given by the recursive relation:

[0118] V 1，K =P(y1|k)·π k

[0119] V t，K =P(y t |k)·max x∈S (a x，k ·V t-1，x )

[0120] V here t，K This is the probability of the most likely state sequence corresponding to the first t observations with a final state of k. The Viterbi path can be obtained by saving a backward pointer to remember the state x used in the second equation. Declare a function Ptr(k, t) that returns the probability of calculating V if t > 1. t，K The value of x used, or k if t = 1, is as follows:

[0121] x T =argmax x∈S (V T，x )

[0122] x t-1 =Ptr(x t ,t)

[0123] The adaptive projection matching method for 2D and 3D maps based on Hidden Markov chains has the following steps:

[0124] Step 101: Initialize the parameter input of the Hidden Markov Projection Prediction Model based on user preferences and projection characteristics, and define the optimal matching projection sequence.

[0125] Where λ = (A, B, Π), the time sequence of a single user browsing {1, 2, ..., T}, and the initial observation sequence of the projection. Projection Library Map projection comprehensive evaluation error The probability ratio ξ of user prediction and projection evaluation characteristics, and the optimal matching projection sequence.

[0126] Step 102: Normalize user preferences and map projection evaluation values, for Make See formula (16).

[0127] Step 103: Calculate the maximum probability among all paths with an initial state of ζ, that is, for any δ1(ζ)=π′ ζ b ζ (o1).

[0128] Among them, the single time sequence of Earth browsing {1, 2, ..., T}, and the observation sequence of projection. Define variable δ t (ζ) represents the maximum probability among all paths with state ζ at time t, where all paths contain elements defined as... The calculation of δ1(ζ) is given by formula (4).

[0129] Step 104: The maximum path probability of the previous time step is initialized to 0, that is, for any... ψ1(ζ)=0.

[0130] For each state Record the previous state of the path with the highest probability. The probability is See the description of formula (6).

[0131] Step 105: Calculate the observation probability of the composite projection characteristics for any have

[0132] The calculation method for the observation probability is the same as the probability of each state occurring at the initial time, as detailed in formula (17).

[0133] Step 106: Iteratively calculate for t = 2, 3, ..., T.

[0134] The above iterative steps are the principle of the Viterbi algorithm, which seeks the Viterbi path-hidden state sequence that is most likely to produce the sequence of observed events.

[0135] Step 107: For t = T-1, T-2, ..., 1, calculate the optimal projection matching path as follows:

[0136] Wherein, the iteration termination state is

[0137] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0138] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for adaptive projection matching of 2D-3D maps based on Hidden Markov Chains, characterized in that, The steps are as follows: Step 101: constructing a hidden Markov projection prediction model based on user preferences and projection characteristics, the hidden Markov projection prediction model taking a map projection category as a hidden state, taking a user browsing position or a viewing angle as an observation state, fusing user preferences for using a projection and projection standard parameters to form a probability of each hidden state occurring at each time, updating an observation probability at each time, constantly iterating, and forming a current optimal projection matching combination; Step 102: initialize the user's single-browsing time sequence {1, 2, …, T}, the initial observation sequence Projection sequence set Map projection comprehensive evaluation error User preference prediction and projection evaluation characteristic probability proportion ξ, the optimally matched projection sequence Step 103: fuse user preference and projection characteristics, calculate the probability value of the initial time hidden state ζ as π ζ ; Step 104: calculating a maximum probability δ1(ζ) in all paths of the hidden state ζ at an initial time; Step 105: Calculate the observation probability b' of the integrated projection characteristic at the initial time instant ζ (o1); Step 106: iteratively compute the maximum probability δ of all paths in state ζ at time t based on the principles of the Viterbi algorithm t (ζ) and the previous state of the most probable path is the probability ψ t (ζ) of the most probable Viterbi path - the sequence of hidden states - that generated the sequence of observed events; Step 107: Calculate the optimal projection matching path 2. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 1, characterized in that, Step 103, fuse the user preference and the projection characteristic, and calculate the probability value π of the initial time state ζ ζ The specific method is that: Six error measurement parameters are used, namely, isotropy I, area A, bending F, skewness S, distance D and boundary cut B, taking the normalized constant N i = 0.51, N a = 0.41, N f = 0.64, N s = 0.60, N d = 0.449, N b = 0.25, then the plane map projection comprehensive evaluation error S e is represented as: For projection library Let the probability of each state occurring at the initial time be Π = {π1, π2, …, πN}, where πn represents the initial state probability based on the projection pn. n}, wherein π i represents the initial state probability based on the projection pn. i The state probability determined by the user preference is defined as π i' , the state probability determined by the projection characteristics is π i” , and π i' The preset proportion is ξ, 0≤ξ≤1, and it is obtained that: π i = ξπ i' + (1 - ξ)π i” wherein, π i” By normalizing the comprehensive evaluation error of the planar map projection, it is obtained that the projection sequence involved in a single browsing map is The comprehensive evaluation error of the planar map projection is The normalized comprehensive evaluation error of the planar map projection is expressed as Π(S) = {π(S1), π(S2), …, π(S k )}; since the comprehensive evaluation error value is inversely proportional to the similarity of the projection, there is: S e1 π(S1) = S e2 π(S2) =... = S ek π(S k ) π(S1) + π(S2) +... + π(S k ) = 1 It is inferred that: By π(S i ) is calculated π i” , and then get: Accordingly, the probability value π of the initial time state ζ = p i is obtained. ζ ​ 3. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 2, characterized in that, Step 104: calculating a maximum probability δ1(ζ) in all paths of the hidden state ζ at an initial time, and the specific method is: δ1(ζ) = π ζ b ζ (o1) where b ζ (o1) denotes the probability of the hidden state being ζ and the observation state being o1.

4. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 3, characterized in that, Step 105: determining a probability ψ1(ζ) of a previous hidden state h of the maximum probability path at the initial time, and the specific method is: ψ1(ζ) = 0.

5. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 4, characterized in that, Step 105, calculate the observation probability b' of the integrated projection characteristic at the initial moment ζ (o1), and the specific method is:

6. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 5, characterized in that, Step 106, based on the principle of Viterbi algorithm, iteratively calculate the maximum probability of all paths of state ζ at time t δ t (ζ) and the previous state of the maximum probability path is The probability ψ t (ζ) of the maximum probability path, the method is as follows: in, The hidden state is indicated by The probability of converting to ζ.

7. The Hidden Markov Chain based 2D-3D map adaptive projection matching method according to claim 6, characterized in that, Step 107: Calculate the optimal projection matching path The specific method is: Let the probability of the optimal path be P * the end point of the optimal path is Then For t = T-1, T-2, …, 1, the optimal path is backtracked: finding an optimal path 8. A system for adaptive projection matching of 2D-3D maps based on hidden Markov chains, characterized in that, The method for adaptively matching a two-dimensional and three-dimensional map based on a hidden Markov chain is implemented. 9.A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the method for adaptively matching a two-dimensional and three-dimensional map based on a hidden Markov chain is implemented. 10.A computer readable storage medium, having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for adaptively matching a two-dimensional and three-dimensional map based on a hidden Markov chain is implemented.

Images