An online visual interaction system and method based on electronic products

Abstract

The present invention relates to an online visual interaction system and method based on electronic products, and solves the technical problem that the interaction instructions and interaction results are presented separately and the interaction data cannot be processed efficiently and accurately. By adopting an interaction instruction input unit, an interaction instruction processing and running unit, and an interaction output unit capable of outputting visual signals; the interaction instruction processing and running unit executes the following steps: step s1, receiving interaction instructions and parsing the interaction instructions; step s2, calling a program for controlling the electronic product to run according to the interaction instructions to obtain a program feedback scenario; step s3, fusing the interaction instruction scenario and the program feedback scenario into a feedback chain virtual scenario, and the feedback chain virtual scenario can simultaneously represent the interaction instructions and the operation feedback results of the electronic product; step s4, outputting the feedback chain virtual scenario after virtual fusion. The technical solution preferably solves this problem and can be used for online visual interaction of electronic products.

Description

Technical Field

[0001] The present invention relates to the field of online interaction of electronic products, and particularly to an online visual interaction system and method based on electronic products. Background Art

[0002] The human-computer interaction system has developed along with the birth of the computer. In modern and future societies, as long as people use information processing technologies such as communication and computers to carry out activities for society, economy, environment, and resources, human-computer interaction is an eternal theme. Given its importance to the development of science and technology, researching how to achieve natural, convenient, and ubiquitous human-computer interaction has become the ultimate goal of modern information technology and artificial intelligence technology research, and is also a new combination point of the intersection of multiple sciences such as mathematics, information science, intelligent science, neuroscience, and physiology and psychology science, and will guide the popular research directions of information and computers in the early 21st century.

[0003] There are problems in the online visual interaction of existing electronic products, such as the separated presentation of interaction instructions and interaction results, and the inability to efficiently and accurately process interaction data. The present invention provides an online visual interaction system and method based on electronic products to solve the above technical problems. Summary of the Invention

[0004] The technical problem to be solved by the present invention is the technical problem existing in the prior art that the interaction instructions and interaction results are presented separately and the interaction data cannot be processed efficiently and accurately. A new online visual interaction system based on electronic products is provided, and the online visual interaction system has the characteristics of well-integrated presentation of interaction instructions and interaction results and being able to efficiently and accurately realize the operation of the system.

[0005] To solve the above technical problems, the following technical solutions are adopted:

[0006] An online visual interaction system based on electronic products, the online visual interaction system based on electronic products includes an interaction instruction input unit, an interaction instruction processing and running unit, and an interaction output unit capable of outputting visual signals;

[0007] The interaction instruction processing and running unit is used to process interaction instructions and control the calling of electronic products according to the interaction instructions. The interaction instruction processing and running unit executes the following steps:

[0008] Step s1, receiving the interaction instruction and parsing the interaction instruction;

[0009] Step s2, calling and controlling the program operation of the electronic product according to the interaction instruction to obtain a program feedback scenario representing the operation feedback result of the electronic product;

[0010] Step s3: Integrate the interactive instruction scenario and the program feedback scenario into a feedback chain virtual scenario, which can simultaneously represent the interactive instructions and the operation feedback results of the electronic product;

[0011] Step s4: Control the interactive output unit to output the feedback chain virtual scenario after virtual integration as the visual feedback result.

[0012] In the above solution, for optimization, further, step s2 also includes preprocessing the interactive instruction data and the electronic product program instruction data. The preprocessing includes:

[0013] Step ss1: Distinguish the data change rates of the interactive instruction data and the electronic product program instruction data. Define the data with a data change rate lower than the preset threshold as high-level data and execute step ss2, and define the rest as low-level data and execute step ss7;

[0014] Step ss2: Invoke the first fusion data processing unit to process the high-level data;

[0015] Step ss3: Define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k is the set of real numbers;

[0016] Step ss4: Calculate the characteristic feedback coefficient ∝ and the divergence coefficient γ through y k = μ + αt k + ε k , where μ = log(2γ); here, ε k is the preset error term coefficient, and t k = log|ν k |;

[0017] Step ss5: Calculate the position parameter δ through z k = δν k + ε k , where z k = arctan(Im(ν k ) / Re(ν k );

[0018] Step ss6: Substitute the characteristic feedback coefficient ∝, the divergence coefficient γ, and the position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝} to calculate the Fourier transform and obtain the probability density function f(x), thus completing the fitting estimation fusion of the high-level data P n ;

[0019] Step ss7, define as the advanced data change value, where A is the real-time data value, is the estimated parameter preset by the historical advanced data samples;

[0020] Step ss7, call the second fusion data processing unit to process the low-level data;

[0021] Step ss8, judge the advanced data fitted and estimated in step ss6 and the low-level data fitted and processed in step ss7. If the advanced data change value is greater than the predetermined value threshold T max or the low-level data change value is greater than the predetermined value threshold T max , then the preprocessing is completed and the preprocessing result is output; otherwise, loop to execute step ss1.

[0022] In this preferred solution, by preferentially considering the efficiency of data processing, the real-time performance of system processing is improved, and data fusion estimation is performed to take into account the accuracy of the system.

[0023] Further, the step ss7 includes:

[0024] Step A, interleave the acquisition of the same low-level data value, and define it as the first time feature sample set and the second time feature sample set;

[0025] Step B, calculate the Gaussian convolution transform function of the first time feature sample set and the second time feature sample set, and calculate the Gaussian feature δ = {00, 01, 10, 11} in two-bit binary dimension; calculate the Gaussian convolution feature in three-bit binary dimension

[0026] Step C, normalize the Gaussian feature in step B to obtain the consistency interval function

[0027] where y is the time feature matrix composed of m first time feature vectors or second time feature vectors, μ is the mean of the time feature matrix, and δ is the variance of the time feature matrix;

[0028] Step D, calculate d1 = [d11, d21,... dc1] as the set of the shortest distances between two feature samples in the first time feature sample set, arrange di1 in descending order, and calculate the distance mean where i1 = 1, 2, 3,... c, and dic is the shortest distance;

[0029] Step E, calculate \(d_2 = [d_{12}, d_{22},..., d_{c2}]\) as the set of the shortest distances between two feature samples in the second time feature sample set, arrange \(d_{i2}\) in descending order, and calculate the distance mean value. where \(i2 = 1, 2, 3,..., c\), and \(d_{ic}\) is the shortest distance.

[0030] Step F, calculate the distance weight value. Define the unified weight value of the samples as \(w = [w_1, w_2,..., w m , and the mean value of the weight is

[0031] Step G, calculate the fusion processing function of the low-level data.

[0032] This preferred solution further provides a unique low-speed data estimation and fusion method, which can balance efficiency and accuracy.

[0033] Furthermore, the said step s3 includes:

[0034] Step s31, taking the virtual image of the electronic product as the starting point, heterotopically define the interactive instruction scene area and the program feedback scene area.

[0035] Step s32, identify the real-time changing objects in the interactive instruction scene area and the program feedback scene area and define them as the target space scene model, and the current feedback chain virtual scene as the source space scene model.

[0036] Step s33, extract the background area from the source space scene model Obtain the contour core point information of the background area , including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point.

[0037] Step s34, construct the source space scene data, and judge whether the current pixel point belongs to the background area If the current pixel point belongs to the background area then execute step s35, otherwise execute s36;

[0038] Step s35, obtain the actual RGB component value of the current pixel point. If the actual RGB component value is less than the preset RGB component threshold, define the actual RGB component value as the small RGB component value, and set the RGB component threshold as the RGB component value; otherwise, set the actual RGB component value as the RGB component value.

[0039] Step s36, set the RGB classification value of the current pixel point to 0.

[0040] Step s37, according to the source space scene data and the background area Regarding the contour core point information, pixels with RGB component values all greater than a predefined threshold are defined as background pixels, and the background region of the source space scene is reconstructed according to the background feature purity reconstruction model;

[0041] Step s38, establish a background feature purity reconstruction model:

[0042]

[0043]

[0044]

[0045] d′ p is the feature purity value of the pixel in the target space scene to-be-fused region that has the same pixel position as the background region Pixels p and q are neighboring pixels to each other, and g p and g q respectively represent the gray values of pixel p and its neighboring pixel q in the target space scene to-be-fused region, σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient. is the 4-neighborhood connectivity of pixel p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel q of pixel p in the background region ;

[0046] Step s39, perform AB partitioning on the background region of the source space scene model based on the contour core points, and calculate the range of feature purity values D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); If the target space scene feature purity value corresponding to pixel p in the background region is outside D AB , then execute step s40, otherwise execute step s41;

[0047] Step s40, judge the size of the target space scene feature purity value corresponding to the position of pixel p in the background region and the maximum feature purity value d max . If the target space scene feature purity value is greater than the maximum feature purity value d max, then define the fusion result at the position where the pixel point p is located as the pixel point parameter of the background area, otherwise define it as the pixel point parameter of the target space scene area;

[0048] Step s41, if there is a background area of pixel points then execute step s42, otherwise execute step s43; where qjs is the number of background sub-blocks;

[0049] Step s42, calculate that the feature purity value of the pixel point p in the background sub-block area is d′ p , and return to execute step s39;

[0050]

[0051] Among them, the two-dimensional plane area range corresponding to the background sub-block area composed of the contour core points A and B is R AB ={x s , y s , w, h | w = x e - x s , h = y e - y s}, the starting abscissa of the two-dimensional plane area is x s = min(x A , x B ), the starting ordinate of the two-dimensional plane area is y s = min(y A , y B ), the ending abscissa of the two-dimensional plane area is x e = max(x A , x B ), the ending ordinate of the two-dimensional plane area is y e = max(y A , y B ), and the pixel point corresponding to the minimum feature purity value of the two-dimensional plane area is p min ;

[0052] Step s43, establish a neighborhood microstructure model, calculate the feature purity value of the pixel point p,, and return to execute step s39; the neighborhood microstructure model is the same as the background feature purity reconstruction model, only change w pq to:

[0053]

[0054] Step s44, obtain the feedback chain virtual scene of real-time fusion according to the position fusion result.

[0055] Preferred solution, provides a fusion of an interactive instruction scene and a feedback scene, and can present a dual effect in real time.

[0056] The present invention also provides an online visual interaction method based on an electronic product. The method is based on the established online visual interaction system for electronic products, and the method includes:

[0057] An interaction instruction input unit, an interaction instruction processing and running unit, and an interaction output unit capable of outputting visual signals;

[0058] Step 1, the interaction instruction input unit collects interaction instructions and transmits them to the interaction instruction processing and running unit;

[0059] Step 2, the interaction instruction processing and running unit receives the interaction instructions and parses the interaction instructions;

[0060] Step 3, according to the result of parsing the interaction instructions, the interaction instruction processing and running unit calls to control the electronic product to implement the instruction program operation, and establishes a program feedback scenario representing the operation feedback result of the electronic product according to the program operation result;

[0061] Step 4, the interaction instruction processing and running unit fuses the interaction instruction scenario and the program feedback scenario into a feedback chain virtual scenario, and the feedback chain virtual scenario can represent both the interaction instructions and the operation feedback result of the electronic product;

[0062] Step 5, control the interaction output unit to output the feedback chain virtual scenario after virtual fusion as a visual feedback result.

[0063] Furthermore, step 3 also includes preprocessing the interaction instruction data and the electronic product program instruction data. The preprocessing includes:

[0064] Step ss1, distinguish the data change rates of the interaction instruction data and the electronic product program instruction data, define the data with a data change rate lower than the preset threshold as high-level data, execute step ss2, and define the rest as low-level data and execute step ss7;

[0065] Step ss2, call the first fusion data processing unit to process the high-level data;

[0066] Step ss3, define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k are the real number sets;

[0067] Step ss4, through y k = μ + αt k + ε k, μ = log(2γ), calculate the characteristic feedback coefficient ∝ and the divergence coefficient γ; where, ε k is a preset error term coefficient, t k = log|ν k |;

[0068] Step ss5, through z k = δν k + ε k , calculate the position parameter δ, where, z k = arctan(Im(ν k ) / Re(ν k );

[0069] Step ss6, substitute the characteristic feedback coefficient ∝, the divergence coefficient γ, and the position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝} to calculate the Fourier transform and obtain the probability density function f(x), completing the fitting estimation fusion of the high-level data P n ;

[0070] Step ss7, define as the high-level data change value, where, A is the real-time data value, is the estimation parameter preset from the historical high-level data samples;

[0071] Step ss7, call the second fusion data processing unit to process the low-level data;

[0072] Step ss8, judge the high-level data obtained by the fitting estimation in step ss6 and the low-level data obtained by the fitting processing in step ss7. If the high-level data change value is greater than the predetermined value threshold T max or the low-level data change value is greater than the predetermined value threshold T max , then complete the preprocessing and output the preprocessing result; otherwise, loop and execute step ss1.

[0073] Furthermore, the step ss7 includes:

[0074] Step A, perform interleaved acquisition on the same low-level data value, defined as the first time feature sample set and the second time feature sample set;

[0075] Step B, calculate the Gaussian convolution transform function of the first time feature sample set and the second time feature sample set, calculate the Gaussian feature δ of two-bit binary dimension = {00, 01, 10, 11}; calculate the Gaussian convolution feature of three-bit binary dimension

[0076] Step C, normalize the Gaussian features in step B to obtain the consistency interval function

[0077] Among them, y is a time feature matrix composed of m first time feature vectors or second time feature vectors, μ is the mean value of the time feature matrix, and δ is the variance of the time feature matrix;

[0078] Step D, calculate d1 = [d11, d21,... dc1] as the set of shortest distances between two feature samples in the first time feature sample set, arrange di1 in descending order, and calculate the distance mean value where i1 = 1, 2, 3,... c, and dic is the shortest distance;

[0079] Step E, calculate d2 = [d12, d22,... dc2] as the set of shortest distances between two feature samples in the second time feature sample set, arrange di2 in descending order, and calculate the distance mean value where i2 = 1, 2, 3,... c, and dic is the shortest distance;

[0080] Step F, calculate the distance weight Define the sample unified weight as w = [w1, w2,..., w m , and the mean value of the weight is

[0081] Step G, calculate the fusion processing function of the low-level data

[0082] Furthermore, the fourth step includes:

[0083] Step s31, taking the virtual image of the electronic product as a reference point, heterotopically define the interactive instruction scene area and the program feedback scene area;

[0084] Step s32, identify the real-time changing objects in the interactive instruction scene area and the program feedback scene area and define them as the target space scene model, and the current feedback chain virtual scene as the source space scene model;

[0085] Step s33, extract the background area from the source space scene model Obtain the contour core point information of the background area including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point;

[0086] Step s34, construct the source space scene data, and judge whether the current pixel point belongs to the background area If the current pixel point belongs to the background area then execute step s35, otherwise execute s36;

[0087] Step s35: Obtain the actual RGB component values of the current pixel. If the actual RGB component values are less than the preset RGB component threshold, define the actual RGB component values as the small RGB component values, and set the RGB component threshold as the RGB component values; otherwise, set the actual RGB component values as the RGB component values.

[0088] Step s36: Set all the RGB classification values of the current pixel to 0.

[0089] Step s37: According to the source space scene data and the contour core point information of the background area define the pixels with RGB component values all greater than the predefined threshold as background pixels, and reconstruct the background area of the source space scene according to the background feature purity reconstruction model.

[0090] Step s38: Establish a background feature purity reconstruction model:

[0091]

[0092]

[0093]

[0094] d′ p is the feature purity value of the pixel in the target space scene area to be fused that has the same pixel position as the pixel in the background area Pixels p and q are neighboring pixels to each other. g p and g q represent the gray values of pixel p and its neighboring point q in the target space scene area to be fused respectively. σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient. is the 4-neighborhood connectivity of pixel p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel q of pixel p in the background area ;

[0095] Step s39: Perform AB partitioning on the background area of the source space scene model based on the contour core points, and calculate the range of feature purity values D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); If the background area The purity value of the target spatial scene feature corresponding to the pixel point p is within D AB If it is outside, step s40 is executed; otherwise, step s41 is executed;

[0096] Step s40: Determine the background area The purity value of the target spatial scene feature corresponding to the position of the pixel point p in it, and the maximum feature purity value d max If the purity value of the target spatial scene feature is greater than the maximum feature purity value d max , the fusion result at the position where the pixel point p is located is defined as the pixel point parameter of the background area; otherwise, it is defined as the pixel point parameter of the target spatial scene area;

[0097] Step s41: If there are pixel points in the background area Then step s42 is executed; otherwise, step s43 is executed; where qjs is the number of background sub - blocks;

[0098] Step s42: Calculate the feature purity value of the pixel point p in the background sub - block area as d′ p Return and execute step s39;

[0099]

[0100] The two - dimensional plane region range corresponding to the background sub - block area composed of the contour core points A and B is R AB ={x s ,y s ,w,h|w = x e -x s ,h = y e -y s}, the starting abscissa of the two - dimensional plane region is x s = min(x A ,x B ), the starting ordinate of the two - dimensional plane region is y s = min(y A ,y B ), the ending abscissa of the two - dimensional plane region is x e = max(x A ,x B ), the ending ordinate of the two - dimensional plane region is y e = max(y A ,y B ), and the pixel point corresponding to the minimum feature purity value of the two - dimensional plane region is p min ;

[0101] Step s43: Establish a neighborhood microstructure model, calculate the feature purity value of pixel point p, and return to execute step s39; the neighborhood microstructure model is the same as the background feature purity reconstruction model, except that w is changed. pq It is:

[0102]

[0103] Step s44: Obtain a real-time fused feedback chain virtual scene according to the position fusion result. Description of the Drawings

[0104] The present invention will be further described below in conjunction with the drawings and embodiments.

[0105] Figure 1 Schematic diagram of an online visualization interaction system based on electronic products.

[0106] Figure 2 Schematic diagram of the interactive instruction processing and running unit program.

[0107] Figure 3 Schematic diagram of the feedback chain virtual scene. Detailed Embodiments

[0108] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0109] Embodiment 1

[0110] This embodiment provides an online visualization interaction system based on electronic products, such as Figure 1 The online visualization interaction system based on electronic products includes an interactive instruction input unit, an interactive instruction processing and running unit, and an interactive output unit capable of outputting visualization signals;

[0111] The interactive instruction processing and running unit is used to process interactive instructions and control the calling of electronic products according to the interactive instructions, such as Figure 2 The interactive instruction processing and running unit executes the following steps:

[0112] Step s1: Receive an interactive instruction and parse the interactive instruction;

[0113] Step s2: Call and control the program operation of the electronic product according to the interactive instruction to obtain a program feedback scene representing the operation feedback result of the electronic product;

[0114] Step s3: Integrate the interactive instruction scene and the program feedback scene into a feedback chain virtual scene, and the feedback chain virtual scene can simultaneously represent the interactive instruction and the operation feedback result of the electronic product, such as Figure 3;

[0115] Step s4, control the interactive output unit to output the virtual scene of the feedback chain after virtual fusion as the visual feedback result.

[0116] In the above solution, for optimization, further, step s2 further includes preprocessing the interactive instruction data and the electronic product program instruction data, and the preprocessing includes:

[0117] Step ss1, distinguish the data change rates of the interactive instruction data and the electronic product program instruction data, define the data with a data change rate lower than the preset threshold as high-level data, execute step ss2, and define the rest as low-level data and execute step ss7;

[0118] Step ss2, call the first fusion data processing unit to process the high-level data;

[0119] Step ss3, define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k are the set of real numbers;

[0120] Step ss4, through y k = μ + αt k + ε k , μ = log(2γ), calculate the characteristic feedback coefficient ∝ and the divergence coefficient γ; where, ε k is the preset error term coefficient, t k = log|ν k |;

[0121] Step ss5, through z k = δν k + ε k , calculate the position parameter δ, where z k = arctan(Im(ν k ) / Re(ν k );

[0122] Step ss6, substitute the characteristic feedback coefficient ∝, the divergence coefficient γ, and the position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝}, calculate the Fourier transform, and obtain the probability density function f(x) to complete the fitting estimation fusion of the high-level data P n ;

[0123] Step ss7, define As a high-level data change value, where A is the real-time data value, is the estimation parameter preset by the historical high-level data samples;

[0124] Step ss7, call the second fusion data processing unit to process the low-level data;

[0125] Step ss8, judge the high-level data fitted and estimated in step ss6 and the low-level data fitted and processed in step ss7. If the high-level data change value is greater than the predetermined value threshold T max or the low-level data change value is greater than the predetermined value threshold T max , then the preprocessing is completed and the preprocessing result is output; otherwise, loop to execute step ss1.

[0126] In this preferred solution, by giving priority to the efficiency of data processing, the real-time performance of system processing is improved, and data fusion estimation is carried out to take into account the accuracy of the system.

[0127] Furthermore, the step ss7 includes:

[0128] Step A, interleave the acquisition of the same low-level data value, defined as the first time feature sample set and the second time feature sample set;

[0129] Step B, calculate the Gaussian convolution transform function of the first time feature sample set and the second time feature sample set, calculate the Gaussian feature δ = {00, 01, 10, 11} in two-bit binary dimension; calculate the Gaussian convolution feature in three-bit binary dimension

[0130] Step C, normalize the Gaussian feature in step B to obtain the consistency interval function

[0131] where y is the time feature matrix composed of m first time feature vectors or second time feature vectors, μ is the mean of the time feature matrix, and δ is the variance of the time feature matrix;

[0132] Step D, calculate that d1 = [d11, d21,... dc1] is the set of the shortest distances between two feature samples in the first time feature sample set, arrange di1 in descending order, and calculate the distance mean where i1 = 1, 2, 3,... c, and dic is the shortest distance;

[0133] Step E, calculate that d2 = [d12, d22,... dc2] is the set of the shortest distances between two feature samples in the second time feature sample set, arrange di2 in descending order, and calculate the distance mean where i2 = 1, 2, 3,... c, and dic is the shortest distance;

[0134] Step F, calculate the distance weight Define the unified weight of the samples as w = [w1, w2,..., w m , and the average weight is

[0135] Step G, calculate the fusion processing function of the low-level data

[0136] This preferred solution further provides a unique low-speed data estimation and fusion method, which can balance efficiency and accuracy.

[0137] Furthermore, the said step s3 includes:

[0138] Step s31, taking the virtual image of the electronic product as the starting point, heterotopically define the interactive instruction scene area and the program feedback scene area;

[0139] Step s32, identify the real-time changing objects in the interactive instruction scene area and the program feedback scene area and define them as the target space scene model, and the current feedback chain virtual scene as the source space scene model;

[0140] Step s33, extract the background area from the source space scene model Obtain the contour core point information of the background area , including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point;

[0141] Step s34, construct the source space scene data, and judge whether the current pixel point belongs to the background area If the current pixel point belongs to the background area Then execute step s35, otherwise execute s36;

[0142] Step s35, obtain the actual RGB component value of the current pixel point. If the actual RGB component value is less than the preset RGB component threshold, then define the actual RGB component value as the small RGB component value, and set the RGB component threshold as the RGB component value; otherwise, set the actual RGB component value as the RGB component value;

[0143] Step s36, set the RGB classification value of the current pixel point to 0;

[0144] Step s37, according to the source space scene data and the contour core point information of the background area Define the pixel points with RGB component values greater than the predefined threshold as background pixel points, and reconstruct the background area of the source space scene according to the background feature purity reconstruction model;

[0145] Step s38, establish a background feature purity reconstruction model:

[0146]

[0147]

[0148]

[0149] d′ p is the feature purity value of the pixel points in the area to be fused in the target space scene that have the same pixel point positions as those in the background area Pixels p and q are neighboring pixel points. g p and g q respectively represent the gray values of pixel point p and its neighboring point q in the area to be fused in the target space scene. σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient. is the 4-neighborhood connectivity of pixel point p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel point q of pixel point p in the background area ;

[0150] Step s39, perform AB partitioning on the background area of the source space scene model based on the contour core points, and calculate the range of the feature purity value as D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); If the feature purity value of the pixel point p corresponding in the background area in the target space scene is outside D AB , then execute step s40, otherwise execute step s41;

[0151] Step s40, judge the size of the feature purity value of the target space scene corresponding to the position of pixel point p in the background area and the maximum feature purity value d max . If the feature purity value of the target space scene is greater than the maximum feature purity value d max , then define the fusion result at the position where pixel point p is located as the pixel point parameter of the background area, otherwise define it as the pixel point parameter of the target space scene area;

[0152] Step s41, if there are pixel points in the background area Then, step S42 is executed; otherwise, step S43 is executed. Here, qjs is the number of background sub-blocks;

[0153] In step S42, the feature purity value d' of the pixel point p in the background sub-block area is calculated p , and then return to execute step S39;

[0154]

[0155] Among them, the two-dimensional plane area range R corresponding to the background sub-block area composed of the contour core points A and B is AB ={x s , y s , w, h | w = x e - x s , h = y e - y s}}. The starting abscissa of the two-dimensional plane area is x s = min(x A , x B ), the starting ordinate of the two-dimensional plane area is y s = min(y A , y B ), the ending abscissa of the two-dimensional plane area is x e = max(x A , x B ), and the ending ordinate of the two-dimensional plane area is y e = max(y A , y B ). The pixel point corresponding to the minimum feature purity value of the two-dimensional plane area is p min ;

[0156] In step S43, a neighborhood micro-structure model is established, the feature purity value of the pixel point p is calculated, and then return to execute step S39; the neighborhood micro-structure model is the same as the background feature purity reconstruction model, except that w pq is changed to:

[0157]

[0158] In step S44, according to the position fusion result, a real-time fused feedback chain virtual scene is obtained.

[0159] In a preferred solution, a fusion of an interaction instruction scene and a feedback scene is provided, which can present a dual effect in real time.

[0160] The present invention also provides an online visualization interaction method based on an electronic product. The method is based on the establishment of the online visualization interaction system of the electronic product described above. The method includes:

[0161] An interactive instruction input unit, an interactive instruction processing and running unit, and an interactive output unit capable of outputting visual signals;

[0162] Step 1: The interactive instruction input unit collects interactive instructions and transmits them to the interactive instruction processing and running unit;

[0163] Step 2: The interactive instruction processing and running unit receives the interactive instructions and parses them;

[0164] Step 3: According to the result of parsing the interactive instructions, the interactive instruction processing and running unit calls to control the electronic product to implement the instruction program operation, and establishes a program feedback scenario representing the operation feedback result of the electronic product according to the program operation result;

[0165] Step 4: The interactive instruction processing and running unit fuses the interactive instruction scenario and the program feedback scenario into a feedback chain virtual scenario, and the feedback chain virtual scenario can represent both the interactive instructions and the operation feedback result of the electronic product;

[0166] Step 5: Control the interactive output unit to output the feedback chain virtual scenario after virtual fusion as a visual feedback result.

[0167] Further, Step 3 also includes preprocessing the interactive instruction data and the electronic product program instruction data, and the preprocessing includes:

[0168] Step ss1: Distinguish the data change rates of the interactive instruction data and the electronic product program instruction data, define the data with a data change rate lower than the preset threshold as high-level data, execute Step ss2, and define the rest as low-level data and execute Step ss7;

[0169] Step ss2: Call the first fusion data processing unit to process the high-level data;

[0170] Step ss3: Define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k is the real number set;

[0171] Step ss4: Calculate the characteristic feedback coefficient ∝ and the divergence coefficient γ through y k = μ + αt k + ε k , where μ = log(2γ); among them, ε k is the preset error term coefficient, t k = log|ν k |;

[0172] Step ss5, through z k = δν k + ε k , calculate the position parameter δ, where z k = arctan(Im(ν k ) / Re(ν k ));

[0173] Step ss6, substitute the feature feedback coefficient ∝, divergence coefficient γ, and position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝} to calculate the Fourier transform and obtain the probability density function f(x), completing the fitting estimation fusion of the high-level data P n ;

[0174] Step ss7, define as the high-level data change value, where A is the real-time data value, is the estimation parameter preset from historical high-level data samples;

[0175] Step ss7, call the second fusion data processing unit to process the low-level data;

[0176] Step ss8, judge the high-level data obtained by the fitting estimation in step ss6 and the low-level data obtained by the fitting processing in step ss7. If the high-level data change value is greater than the predetermined value threshold T max or the low-level data change value is greater than the predetermined value threshold T max , then complete the preprocessing and output the preprocessing result; otherwise, loop to execute step ss1.

[0177] Furthermore, the step ss7 includes:

[0178] Step A, perform interleaved acquisition on the same low-level data value, defined as the first-time feature sample set and the second-time feature sample set;

[0179] Step B, calculate the Gaussian convolution transform function of the first-time feature sample set and the second-time feature sample set, and calculate the Gaussian features of two-bit binary dimensions δ = {00, 01, 10, 11}; calculate the Gaussian convolution features of three-bit binary dimensions

[0180] Step C, normalize the Gaussian features in step B to obtain the consistency interval function

[0181] where y is the time feature matrix composed of m first-time feature vectors or second-time feature vectors, μ is the mean of the time feature matrix, and δ is the variance of the time feature matrix;

[0182] Step D: Calculate \(d1 = [d11, d21,...dc1]\) as the set of the shortest distances between two feature samples in the first time feature sample set. Arrange \(di1\) in descending order and calculate the distance mean value. where \(i1 = 1, 2, 3,...c\) and \(dic\) is the shortest distance.

[0183] Step E: Calculate \(d2 = [d12, d22,...dc2]\) as the set of the shortest distances between two feature samples in the second time feature sample set. Arrange \(di2\) in descending order and calculate the distance mean value. where \(i2 = 1, 2, 3,...c\) and \(dic\) is the shortest distance.

[0184] Step F: Calculate the distance weight value. Define the sample unified weight as \(w = [w1, w2,..., w m \), and the mean value of the weights is

[0185] Step G: Calculate the fusion processing function of the low-level data.

[0186] Furthermore, the fourth step includes:

[0187] Step s31: Taking the virtual image of the electronic product as the starting point, define the interactive instruction scenario area and the program feedback scenario area ectopically.

[0188] Step s32: Identify the real-time changing objects in the interactive instruction scenario area and the program feedback scenario area and define them as the target space scenario model, and the current feedback chain virtual scenario as the source space scenario model.

[0189] Step s33: Extract the background area from the source space scenario model Obtain the contour core point information of the background area, including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point.

[0190] Step s34: Construct the source space scenario data and determine whether the current pixel point belongs to the background area If the current pixel point belongs to the background area then execute step s35, otherwise execute s36;

[0191] Step s35: Obtain the actual RGB component value of the current pixel point. If the actual RGB component value is less than the preset RGB component threshold, define the actual RGB component value as the small RGB component value and set the RGB component threshold as the RGB component value; otherwise, set the actual RGB component value as the RGB component value.

[0192] Step s36, set the RGB classification values of the current pixel to 0;

[0193] Step s37, based on the source space scene data and the contour core point information of the background area define the pixels with RGB component values greater than the predefined threshold as background pixels, and reconstruct the background area of the source space scene according to the background feature purity reconstruction model;

[0194] Step s38, establish the background feature purity reconstruction model:

[0195]

[0196]

[0197]

[0198] d′ p is the feature purity value of the pixel in the area to be fused in the target space scene that has the same pixel position as the pixel in the background area Pixels p and q are neighboring pixels to each other. g p and g q represent the gray values of pixel p and its neighboring pixel q in the area to be fused in the target space scene respectively. σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient. is the 4-neighborhood connectivity of pixel p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel q of pixel p in the background area ;

[0199] Step s39, divide the background area of the source space scene model into AB blocks based on the contour core points, and calculate the range of the feature purity value D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); If the target space scene feature purity value corresponding to pixel p in the background area is outside D AB , then execute step s40, otherwise execute step s41;

[0200] Step s40, judge the background area The purity value of the target space scene feature corresponding to the position of pixel p, and the maximum feature purity value d max If the purity value of the target space scene feature is greater than the maximum feature purity value d max , then define the fusion result at the position of pixel p as the pixel point parameter of the background area, otherwise define it as the pixel point parameter of the target space scene area;

[0201] Step s41, if there is a background area Pixels Then execute step s42, otherwise execute step s43; where qjs is the number of background sub-blocks;

[0202] Step s42, calculate the feature purity value of pixel p in the background sub-block area as d′ p , and return to execute step s39;

[0203]

[0204] Among them, the two-dimensional plane area range corresponding to the background sub-block area composed of the contour core points A and B is R AB ={x s ,y s ,w,h|w = x e -x s , h = y e -y s}}, the starting abscissa of the two-dimensional plane area is x s = min(x A ,x B ), the starting ordinate of the two-dimensional plane area is y s = min(y A ,y B ), the ending abscissa of the two-dimensional plane area is x e = max(x A ,x B ), the ending ordinate of the two-dimensional plane area is y e = max(y A ,y B ), and the pixel corresponding to the minimum feature purity value of the two-dimensional plane area is p min ;

[0205] Step s43, establish a neighborhood microstructure model, calculate the feature purity value of pixel p,, and return to execute step s39; the neighborhood microstructure model is the same as the background feature purity reconstruction model, only change w pq To:

[0206]

[0207] Step s44: Obtain a feedback chain virtual scene with real-time fusion according to the position fusion result.

[0208] For parts not described in this embodiment, existing technologies can be adopted for implementation, and details are not elaborated in this embodiment.

[0209] Although the illustrative specific embodiments of the present invention are described above for those skilled in the art to understand the present invention, the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, all inventions made using the concept of the present invention are within the scope of protection.

Claims

1. An online visual interaction system based on electronic products, characterized in that: The online visual interaction system based on electronic products includes an interaction instruction input unit, an interaction instruction processing and running unit, and an interaction output unit capable of outputting visual signals; The interaction instruction processing and running unit is used to process interaction instructions, control and call electronic products according to the interaction instructions. The interaction instruction processing and running unit executes the following steps: Step s1, receive the interaction instruction and parse the interaction instruction; Step s2, call and control the program running of the electronic product according to the interaction instruction to obtain a program feedback scenario representing the operation feedback result of the electronic product; Step s3, fuse the interaction instruction scenario and the program feedback scenario into a feedback chain virtual scenario, and the feedback chain virtual scenario can represent both the interaction instruction and the operation feedback result of the electronic product; Step s4, control the interaction output unit to output the feedback chain virtual scenario after virtual fusion as a visual feedback result; The step s3 includes: Step s31, taking the virtual image of the electronic product as a point, defining the interaction instruction scenario area and the program feedback scenario area ectopically; Step s32, identifying the real-time changing objects in the interaction instruction scenario area and the program feedback scenario area and defining them as the target space scenario model, and the current feedback chain virtual scenario as the source space scenario model; Step s33: Extract the background region from the source space scene model Obtain the background region and the contour core point information, including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point; Step s34, construct source space scene data, and determine whether the current pixel point belongs to the background area If the current pixel point belongs to the background area Then execute step s35, otherwise execute s36; Step s35, obtain the actual RGB component value of the current pixel point. If the actual RGB component value is less than the preset RGB component threshold, define the actual RGB component value as the small RGB component value, and set the RGB component threshold as the RGB component value; otherwise, set the actual RGB component value as the RGB component value; Step s36, set the RGB classification value of the current pixel point to 0; Step s37, according to the source space scene data and the contour core point information of the background area define the pixel points with RGB component values all greater than the predefined threshold as background pixel points, and reconstruct the background area of the source space scene according to the background feature purity reconstruction model; Step s38, establish a background feature purity reconstruction model: d′ p is the feature purity value of the pixel points in the area to be fused in the target space scene that have the same pixel point positions as those in the background area ; the pixel point p and the pixel point q are neighboring pixel points of each other, and g p and g q respectively represent the gray values of the pixel point p and its neighboring point q in the area to be fused in the target space scene; σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient is the 4-neighborhood connectivity of the pixel point p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel point q of the pixel point p in the background area ; Step s39, the background area of the source space scene model is AB-blocked based on the contour core points, and the range of the calculated feature purity value is D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); if the target space scene feature purity value corresponding to the pixel point p in the background area is outside D AB , then step s40 is executed, otherwise step s41 is executed; Step s40, determine the background region of the purity value of the target spatial scene feature corresponding to the position of the pixel point p in max and the size of the maximum feature purity value d max If the purity value of the target spatial scene feature is greater than the maximum feature purity value d max , then define the fusion result at the position where the pixel point p is located as the pixel point parameter of the background region, otherwise define it as the pixel point parameter of the target spatial scene region; Step s41, if there is a background area pixel points then execute step s42, otherwise execute step s43; where qjs is the number of background sub-blocks; Step s42, calculate that the feature purity value of the pixel point p in the background sub-block area is d', p and return to execute step s39; Among them, the two-dimensional plane region range corresponding to the background sub-block area composed of the contour core points A and B is R AB ={x s ,y s ,w,h|w = x e -x s ,h = y e -y s}}, the starting abscissa of the two-dimensional plane region is x s = min(x A ,x B ), the starting ordinate of the two-dimensional plane region is y s = min(y A ,y B ), the ending abscissa of the two-dimensional plane region is x e = max(x A ,x B ), the ending ordinate of the two-dimensional plane region is y e = max(y A ,y B ), and the pixel point corresponding to the minimum feature purity value of the two-dimensional plane region is p min ; Step s43: Establish a neighborhood microstructure model, calculate the feature purity value of pixel point p, and return to execute step s39; the neighborhood microstructure model is consistent with the background feature purity reconstruction model, only changing w pq is as follows: Step s44, obtain the feedback chain virtual scenario of real-time fusion according to the position fusion result.

2. The online visualization interaction system based on electronic products according to claim 1, wherein: Step s2 also includes preprocessing the interaction instruction data and the electronic product program instruction data. The preprocessing includes: Step ss1, distinguish the data change rates of the interaction instruction data and the electronic product program instruction data, define the data with a data change rate lower than the preset threshold as high-level data, execute step ss2, and define the rest as low-level data, execute step ss7; Step ss2, call the first fusion data processing unit to process the high-level data; Step ss3, define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k is the set of real numbers; Step ss4, through y k = μ + αt k + ε k , μ = log(2γ), calculate the characteristic feedback coefficient ∝ and the divergence coefficient γ; where ε k is the preset error term coefficient, t k = log|ν k |; Step ss5, through z k = δν k + ε k , calculate the position parameter δ, where z k = arctan(Im(ν k ) / Re(ν k ); Step ss6, substitute the feature feedback coefficient ∝, divergence coefficient γ, and position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝}, calculate the Fourier transform to obtain the probability density function f(x), and complete the fitting estimation fusion of the high-level data P n ; Step ss7, define as the advanced data change value, where A is the real-time data value, and is the estimated parameter preset by the historical advanced data samples; Step ss7, call the second fusion data processing unit to process the low-level data; Step ss8: Determine the high-level data estimated by fitting in step ss6 and the low-level data processed by fitting in step ss7. If the change value of the high-level data is greater than the predetermined threshold value T max or the change value of the low-level data is greater than the predetermined threshold value T max , then the preprocessing is completed and the preprocessing result is output; otherwise, loop to execute step ss1.

3. The online visual interaction system based on electronic products according to claim 2, characterized in that: The step ss7 includes: Step A, perform interleaved acquisition on the same low-level data value, and define it as the first time feature sample set and the second time feature sample set; Step B, calculate the Gaussian convolution transform function of the first time feature sample set and the second time feature sample set, and calculate the Gaussian features δ = {00, 01, 10, 11} in two-bit binary dimension; calculate the Gaussian convolution features in three-bit binary dimension Step C, normalize the Gaussian features in step B to obtain a consistency interval function Wherein, y is a time feature matrix composed of m first time feature vectors or second time feature vectors, μ is the mean value of the time feature matrix, and δ is the variance of the time feature matrix; Step D: Calculate \(d1 = [d11, d21,...dc1]\) as the set of the shortest distances between two feature samples in the first time feature sample set. Arrange \(di1\) in descending order and calculate the mean distance. where \(i1 = 1, 2, 3,...c\) and \(dic\) is the shortest distance. Step E, calculate \(d_2 = [d_{12}, d_{22},..., d_{c2}]\) as the set of shortest distances between two feature samples in the second time feature sample set, sort \(d_{i2}\) in descending order, and calculate the distance mean. where \(i2 = 1, 2, 3,..., c\), and \(d_{ic}\) is the shortest distance. Step F, calculate the distance weight Define the unified weight of the samples as w = [w1, w2,..., w m , and the average weight is Step G, calculate the fusion processing function of the low-level data 4. An online visual interaction method based on electronic products, characterized in that: The method is based on the online visual interaction system for electronic products according to any one of claims 1-3. The method includes: An interaction instruction input unit, an interaction instruction processing and running unit, and an interaction output unit capable of outputting visual signals; Step one, the interaction instruction input unit collects the interaction instruction and transmits it to the interaction instruction processing and running unit; Step two, the interaction instruction processing and running unit receives the interaction instruction and parses the interaction instruction; Step 3: Based on the result of parsing the interaction instruction, the interaction instruction processing and running unit calls to control the electronic product to implement the instruction program operation, and establishes a program feedback scenario representing the operation feedback result of the electronic product according to the program operation result; Step 4: The interaction instruction processing and running unit fuses the interaction instruction scenario and the program feedback scenario into a feedback chain virtual scenario, which can represent both the interaction instruction and the operation feedback result of the electronic product; Step 5: Control the interaction output unit to output the feedback chain virtual scenario after virtual fusion as the visual feedback result.

5. The online visualization interaction method based on electronic products according to claim 4, characterized in that: Step 3 also includes preprocessing the interaction instruction data and the electronic product program instruction data. The preprocessing includes: Step ss1: Distinguish the data change rates of the interaction instruction data and the electronic product program instruction data. Define the data with a data change rate lower than the preset threshold as high-level data, and execute Step ss2. Define the rest as low-level data and execute Step ss7; Step ss2: Call the first fusion data processing unit to process the high-level data; Step ss3, define where k is a positive integer less than or equal to K, and K is the number of independent high-level data in the high-level database, {x1, x2,... x k , x K} are the measured values of K data, j and ν are preset parameters, and ν1, ν2,... ν k is the set of real numbers; Step ss4, through y k = μ + αt k + ε k , μ = log(2γ), calculate the feature feedback coefficient ∝ and the divergence coefficient γ; where ε k is the preset error term coefficient, t k = log|ν k |; Step ss5, through z k = δν k + ε k , calculate the position parameter δ, where z k = arctan(Im(ν k ) / Re(ν k ); Step ss6, substitute the feature feedback coefficient ∝, divergence coefficient γ, and position parameter δ into φ(ν) = exp{jδν - γ|ν| ∝}, calculate the Fourier transform to obtain the probability density function f(x), and complete the fitting estimation fusion of the high-level data P n ; Step ss7, define as the advanced data change value, where A is the real-time data value, and is the estimation parameter preset by the historical advanced data samples; Step ss7: Call the second fusion data processing unit to process the low-level data; Step ss8, determine the high-level data estimated by fitting in step ss6 and the low-level data processed by fitting in step ss7. If the change value of the high-level data is greater than the predetermined threshold T max or the change value of the low-level data is greater than the predetermined threshold T max , then the preprocessing is completed and the preprocessing result is output; otherwise, loop to execute step ss1.

6. The online visualization interaction method based on electronic products according to claim 5, wherein: The said Step ss7 includes: Step A: Interleaved acquisition of the same low-level data value, defined as the first time feature sample set and the second time feature sample set; Step B, calculate the Gaussian convolution transform function of the first time feature sample set and the second time feature sample set, and calculate the Gaussian features δ = {00, 01, 10, 11} in two-bit binary dimension; calculate the Gaussian convolution features in three-bit binary dimension Step C: Normalize the Gaussian features in Step B to obtain a consistency interval function Among them, y is a time feature matrix composed of m first time feature vectors or second time feature vectors, μ is the mean of the time feature matrix, and δ is the variance of the time feature matrix; Step D, calculate that d1 = [d11, d21,...dc1] is the set of shortest distances between two feature samples in the first time feature sample set, arrange di1 in descending order, and calculate the distance mean value. where i1 = 1, 2, 3,...c, and dic is the shortest distance; Step E, calculate \(d_2 = [d_{12}, d_{22},... d_{c2}]\) as the set of the shortest distances between two feature samples in the second time feature sample set, arrange \(d_{i2}\) in descending order, and calculate the distance mean value. where \(i2 = 1, 2, 3,... c\), and \(d_{ic}\) is the shortest distance. Step F, calculate the distance weight Define the unified weight of the samples as w = [w1, w2,..., w m , and the mean value of the weights is Step G, calculate the fusion processing function of the low-level data 7. The online visual interaction system based on electronic products according to claim 5, wherein: The said Step 4 includes: Step s31: Taking the virtual image of the electronic product as the starting point, define the interaction instruction scenario area and the program feedback scenario area ectopically; Step s32: Identify the real-time changing objects in the interaction instruction scenario area and the program feedback scenario area and define them as the target space scenario model, and the current feedback chain virtual scenario as the source space scenario model; Step s33: Extract the background region from the source space scene model Obtain the background region and the contour core point information thereof, including the feature purity value of the core point and the two-dimensional plane coordinate value of the core point; Step s34: Construct source space scene data and determine whether the current pixel belongs to the background area If the current pixel belongs to the background area Then execute step s35; otherwise, execute s36 Step s35: Obtain the actual RGB component value of the current pixel point. If the actual RGB component value is less than the preset RGB component threshold, define the actual RGB component value as the small RGB component value, and set the RGB component threshold as the RGB component value; otherwise, set the actual RGB component value as the RGB component value; Step s36: Set the RGB classification value of the current pixel point to 0; Step s37, based on the source space scene data and the contour core point information of the background area define the pixel points with RGB component values all greater than the predefined threshold as background pixel points, and reconstruct the background area of the source space scene according to the background feature purity reconstruction model; Step s38: Establish a background feature purity reconstruction model: d′ p is the feature purity value of the pixel points in the area to be fused in the target space scene that have the same pixel point positions as those in the background area where the pixel point p and the pixel point q are neighboring pixel points, and g p and g q respectively represent the gray values of the pixel point p and its neighboring point q in the area to be fused in the target space scene, σ g is the standard deviation based on the Gaussian function, and t is a constant coefficient is the 4-neighborhood connectivity of the pixel point p, and d′ q (t - 1) is the known feature purity value of the neighboring pixel point q of the pixel point p in the background area ; Step s39, for the background area of the source space scene model perform AB partitioning based on the contour core points, and calculate that the range of the feature purity value is D AB ={d|d min ≤d≤d max}, the minimum feature purity value d min =min(d A ,d B ), the maximum feature purity value d max =max(d A ,d B ); if the target space scene feature purity value corresponding to the pixel point p in the background area is outside D AB , then execute step s40, otherwise execute step s41; Step s40, determine the background region The purity value of the target space scene feature corresponding to the position of the pixel point p in max is compared with the size of the maximum feature purity value d max . If the purity value of the target space scene feature is greater than the maximum feature purity value d , then define the fusion result at the position where the pixel point p is located as the pixel point parameter of the background region, otherwise define it as the pixel point parameter of the target space scene region; Step s41, if there is a background area of pixel points then execute step s42, otherwise execute step s43; where qjs is the number of background sub - blocks; Step s42, calculate that the feature purity value of the pixel point p in the background sub-block area is d', p , return to execute step s39; Among them, the two-dimensional plane region range corresponding to the background sub-block area composed of the contour core points A and B is R AB ={x s ,y s ,w,h|w = x e - x s ,h = y e - y s}}, the starting abscissa of the two-dimensional plane region is x s = min(x A ,x B ), the starting ordinate of the two-dimensional plane region is y s = min(y A ,y B ), the ending abscissa of the two-dimensional plane region is x e = max(x A ,x B ), the ending ordinate of the two-dimensional plane region is y e = max(y A ,y B ), and the pixel point corresponding to the minimum feature purity value of the two-dimensional plane region is p min ; Step s43, establish a neighborhood microstructure model, calculate the feature purity value of pixel p, and return to execute step s39; the neighborhood microstructure model is the same as the background feature purity reconstruction model, except that w is changed pq is as follows: Step s44: Obtain the feedback chain virtual scenario of real-time fusion according to the position fusion result.

Images