A target recognition result ground display method suitable for image terminal guided ammunition

Abstract

The present application belongs to the technical field of weapon ammunition, and particularly relates to a target recognition result ground display method suitable for image terminal guided ammunition. For the image terminal guided ammunition using target ground assisted recognition and the target confirmation performed by the operator, in order to more accurately and intuitively display the recognition result, after the ground operating display device obtains the recognition result of the target, the target comprehensive confidence degree representing the stable recognition degree in a period of time is obtained by constructing a recognition confidence weight function, and the numerical parameters of the target comprehensive confidence degree, the continuous recognition time length and other recognition results are converted into image display parameters which are more convenient to observe and understand, and are displayed through the prompt box of the ground operating display device, so as to facilitate the target identification and confirmation of the operator. The scheme is simple to implement and effective in method, can be popularized to other applications using machine assisted recognition and human in the loop confirmation, and has certain popularization and application space.

Description

Technical Field

[0001] This invention belongs to the field of weapon and ammunition technology, specifically relating to a ground display method for target identification results of image-guided munitions. This technology can be applied to ground display of identification results of image-guided munitions. Background Technology

[0002] With the rapid development and widespread application of image recognition technology, image-guided terminal guidance munitions have gradually evolved from the traditional "human-in-the-loop" full-process control to a new mode based on target ground-assisted identification for target confirmation and autonomous tracking by the munition's seeker. Compared to the traditional "human-in-the-loop" full-process control mode, this mode requires operators to identify and confirm targets only based on the results provided by the target assistance identification, eliminating the need for continuous operation of the tracking gate. This reduces reliance on human intervention and significantly improves the success rate of target recognition under high terminal velocities and complex background conditions, representing an important future development direction for image-guided terminal guidance.

[0003] The current method for displaying recognition results typically involves overlaying the target onto the image using a prompt box based on the target's coordinates and pixel size, with a target number added next to the prompt box. Operators then identify and confirm the target based on the prompt box. However, due to the limited range of the seeker and the need to control the convergence period to ensure accuracy after switching to terminal guidance, operators often need to complete target identification and confirmation within a few seconds. When multiple recognition results exist in the image, operators must sequentially and indiscriminately identify and select the appropriate result, which presents significant challenges. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] The technical problem to be solved by this invention is: for image-guided munitions that employ ground-assisted target identification, human-in-the-loop target confirmation, and autonomous tracking by the seeker, how to obtain target confidence information that characterizes the degree of stable identification over a period of time after the ground control and display device receives the target ground-assisted identification result, and how to convert the numerical parameters of the identified target into image display parameters that are easier to observe and understand, so as to reduce the difficulty for operators to identify multiple identification results in a short period of time and improve the target identification success rate.

[0006] (II) Technical Solution

[0007] To address the aforementioned technical problems, this invention provides a ground-based method for displaying target identification results of image-guided munitions, the method comprising the following steps:

[0008] Step S1: The ground control and display device receives the target ground-assisted identification result of the missile-borne image. The target ground-assisted identification result includes the number of all identified targets in the image, identification confidence level, continuous identification duration, target centroid coordinates and pixel size information.

[0009] Step S2: Construct a confidence weight function with time as the independent variable;

[0010] Step S3: Calculate the comprehensive confidence level of the target by using the confidence weight function for all identified targets.

[0011] Step S4: Establish a conversion function from numerical parameters of the recognition result to image display parameters. Take the target comprehensive confidence and continuous recognition time as the function input, and select image display parameters including the color parameter value of the target prompt box and the line width value as the function output.

[0012] Step S5: The ground control and display device overlays all identified target information onto the image screen transmitted from the missile based on the calculation results of the conversion function, so as to help the operator identify and confirm the identified targets.

[0013] In step S1, the confidence level of target i is represented by W. i The confidence threshold is set to 0.3, representing the degree of matching of the identification results. Therefore, the identification confidence level satisfies W. i ∈[0.3, 1].

[0014] In step S1, the duration of continuous identification of target i is denoted as T. i The centroid coordinates of target i are represented by the horizontal and vertical pixel values ​​of its distance from the top-left corner of the image as (Nx i ,Ny i The pixel size of target i is represented by its horizontal and vertical pixel values ​​as (Lx i ,Ly i ).

[0015] In step S2, a confidence weight function with time as the independent variable is constructed.

[0016] The duration of the determination period is denoted as the weighted time T. d Select the one that generally satisfies T d The weighting function is defined in the range [0.5, 3], with time t, where t ∈ [0, T]. d [ ] is the independent variable, and the function form is as follows:

[0017]

[0018] In equation (1), Q(t) is the weight value corresponding to time t, and the physical meaning of t is the time length from the current moment.

[0019] In step S3, a confidence weighting function is used to calculate the comprehensive confidence score of all identified targets to characterize the confidence level of the targets within the judgment period; specifically, this is performed as follows:

[0020] Step S31: Based on the weighted time T d The number of recognitions N within the range generates a set of discrete time series values ​​T. w (k), the value of which is given in expression (2):

[0021] T w (k)=k·T s k = 1, 2, ..., N (2)

[0022] Among them, T s T is the identification cycle for ground-assisted target identification. s Typically, it lasts for tens of milliseconds, T w (k) represents the recognition time corresponding to the kth frame previously;

[0023] Step S32: Generate a set of data corresponding to the discrete time series T w (k) Target i identification confidence sequence W i (k) is used to represent the weighted time T. d Recognition confidence level within the range;

[0024] Among them, W i (k) represents the identification confidence value of the k-th frame of the received target i, where k = 1, 2, ..., N;

[0025] When the continuous identification time of target i is T i ≥T d Time, that is, the number of times target i is identified N within the time period. i =N, directly take the corresponding N frames within the time range to identify the confidence level to form a sequence W. i ;

[0026] When the continuous identification time of target i is T i <T d Time, that is, the number of times target i is identified N within the time period. i <N, take the first N i The recognition confidence scores corresponding to the frames form a sequence W. i (k)Previous N i For the first item, set the value of all other items in the sequence to the lower threshold of 0.3, that is:

[0027] W i (j)=0.3 j=Ni N i +1,...,N (3)

[0028] Step S33: Calculate the recognition confidence sequence W for target i. i The weighted average of these values ​​is the overall confidence level ω of the target. i The calculation expression is shown in equation (4):

[0029]

[0030] In step S31, T s It typically lasts for tens of milliseconds.

[0031] In step S4, a conversion function is established to transform the numerical parameters of the recognition result into image display parameters, and the target comprehensive confidence ω is used. i Continuous recognition duration T i As input to the function, select the color parameter value Y of the target tooltip. i Line width value Z i The image display parameters are used as the function output, and the conversion function expression is as shown in equation (5):

[0032]

[0033] Step S4 includes:

[0034] Step S41, the color parameter Y of the target tooltip. i Select to display RGB values, denoted as Y. i =(R i G i B i ), and set the function to the target comprehensive confidence level ω. i Single input, confidence ω i The larger the value, the more vibrant and eye-catching the displayed color will be. The function form is as shown in expression (6):

[0035]

[0036] Step S42: Set the line width value Z i The function is the continuous recognition duration T. i Single input, duration T i The larger the value, the larger the line width value. The function form is as shown in expression (7):

[0037]

[0038] In equation (7), T min This is the lower limit of the time threshold, typically set to 0.5 seconds; T max This is the upper limit of the time threshold, generally related to the weighted time T.d Take the same value.

[0039] The method described herein obtains numerical parameters that characterize the "similarity" of targets by constructing a confidence weight function, and converts the recognition results into image display parameters that are easier to observe and understand. These parameters are then displayed through prompt boxes on the ground control display device, which can effectively show the "similarity" information of the recognition results and help operators to identify multiple targets in sequence according to the prompts, thereby improving the success rate of target recognition.

[0040] (III) Beneficial Effects

[0041] Compared with existing technologies, the method of this invention obtains numerical parameters that characterize the "similarity" of targets by constructing a confidence weight function, and converts the recognition results into image display parameters that are easier to observe and understand. These parameters are displayed through prompts on a ground-based control device, effectively showcasing the "similarity" information of the recognition results and helping operators to sequentially identify multiple targets based on the prompts, thus improving the target recognition success rate. This solution is simple to implement and effective, and can be extended to other applications such as machine-assisted recognition and human-in-the-loop verification, demonstrating considerable potential for widespread application. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the weapon system's composition and information flow.

[0043] Figure 2 Flowchart of the method for displaying target recognition results on the ground. Detailed Implementation

[0044] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0045] To address the aforementioned technical problems, this invention provides a ground-based method for displaying target identification results of image-guided munitions, the method comprising the following steps:

[0046] Step S1: The ground control and display device receives the target ground-assisted identification result of the missile-borne image. The target ground-assisted identification result includes the number of all identified targets in the image, identification confidence level, continuous identification duration, target centroid coordinates and pixel size information.

[0047] Step S2: Construct a confidence weight function with time as the independent variable;

[0048] Step S3: Calculate the comprehensive confidence level of the target by using the confidence weight function for all identified targets.

[0049] Step S4: Establish a conversion function from numerical parameters of the recognition result to image display parameters. Take the target comprehensive confidence and continuous recognition time as the function input, and select image display parameters including the color parameter value of the target prompt box and the line width value as the function output.

[0050] Step S5: The ground control and display device overlays all identified target information onto the image screen transmitted from the missile based on the calculation results of the conversion function, so as to help the operator identify and confirm the identified targets.

[0051] In step S1, the confidence level of target i is represented by W. i The confidence threshold is set to 0.3, representing the degree of matching of the identification results. Therefore, the identification confidence level satisfies W. i ∈[0.3, 1].

[0052] In step S1, the duration of continuous identification of target i is denoted as T. i The centroid coordinates of target i are represented by the horizontal and vertical pixel values ​​of its distance from the top-left corner of the image as (Nx i ,Ny i The pixel size of target i is represented by its horizontal and vertical pixel values ​​as (Lx i ,Ly i ).

[0053] In step S2, a confidence weight function with time as the independent variable is constructed.

[0054] The duration of the determination period is denoted as the weighted time T. d Select the one that generally satisfies T d The weighting function is defined in the range [0.5, 3], with time t, where t ∈ [0, T]. d [ ] is the independent variable, and the function form is as follows:

[0055]

[0056] In equation (1), Q(t) is the weight value corresponding to time t, and the physical meaning of t is the time length from the current moment.

[0057] In step S3, a confidence weighting function is used to calculate the comprehensive confidence score of all identified targets to characterize the confidence level of the targets within the judgment period; specifically, this is performed as follows:

[0058] Step S31: Based on the weighted time T d The number of recognitions N within the range generates a set of discrete time series values ​​T. w (k), the value of which is given in expression (2):

[0059] T w (k)=k·T s k = 1, 2, ..., N (2)

[0060] Among them, T s T is the identification cycle for ground-assisted target identification. s Typically, it lasts for tens of milliseconds, T w (k) represents the recognition time corresponding to the kth frame previously;

[0061] Step S32: Generate a set of data corresponding to the discrete time series T w (k) Target i identification confidence sequence W i (k) is used to represent the weighted time T. d Recognition confidence level within the range;

[0062] Among them, W i (k) represents the identification confidence value of the k-th frame of the received target i, where k = 1, 2, ..., N;

[0063] When the continuous identification time of target i is T i ≥T d Time, that is, the number of times target i is identified N within the time period. i =N, directly take the corresponding N frames within the time range to identify the confidence level to form a sequence W. i ;

[0064] When the continuous identification time of target i is T i <T d Time, that is, the number of times target i is identified N within the time period. i <N, take the first N i The recognition confidence scores corresponding to the frames form a sequence W. i (k)Previous N i For the first item, set the value of all other items in the sequence to the lower threshold of 0.3, that is:

[0065] W i (j)=0.3 j=N i N i +1,...,N (3)

[0066] Step S33: Calculate the recognition confidence sequence W for target i. i The weighted average of these values ​​is the overall confidence level ω of the target. i The calculation expression is shown in equation (4):

[0067]

[0068] In step S31, T s It typically lasts for tens of milliseconds.

[0069] In step S4, a conversion function is established to transform the numerical parameters of the recognition result into image display parameters, and the target comprehensive confidence ω is used. i Continuous recognition duration T i As input to the function, select the color parameter value Y of the target tooltip. i Line width value Z i The image display parameters are used as the function output, and the conversion function expression is as shown in equation (5):

[0070]

[0071] Step S4 includes:

[0072] Step S41, the color parameter Y of the target tooltip. i Select to display RGB values, denoted as Y. i =(R i G i B i ), and set the function to the target comprehensive confidence level ω. i Single input, confidence ω i The larger the value, the more vibrant and eye-catching the displayed color will be. The function form is as shown in expression (6):

[0073]

[0074] Step S42: Set the line width value Z i The function is the continuous recognition duration T. i Single input, duration T i The larger the value, the larger the line width value. The function form is as shown in expression (7):

[0075]

[0076] In equation (7), T min This is the lower limit of the time threshold, typically set to 0.5 seconds; T max This is the upper limit of the time threshold, generally related to the weighted time T. d Take the same value.

[0077] The method obtains numerical parameters that characterize the "similarity" of targets by constructing a confidence weight function, and converts the recognition results into image display parameters that are easier to observe and understand. The results are displayed through prompt boxes on the ground control display device, which can effectively show the "similarity" information of the recognition results and help operators to identify multiple targets in sequence according to the prompts, thereby improving the success rate of target recognition.

[0078] Example 1

[0079] To address the aforementioned technical problems, this invention proposes a ground-based method for displaying target recognition results in image-guided terminal munition weapon systems that employ ground-assisted target identification, human-in-the-loop target confirmation, and autonomous seeker tracking.

[0080] This invention applies to a typical weapon system composition and information flow diagram, see below. Figure 1 The flowchart of the main implementation steps is shown below. Figure 2 The present invention specifically includes the following steps:

[0081] S1. The ground control and display device receives the target ground-aided identification results of the missile-borne image. The identification results include information such as the number of all identified targets in the image, identification confidence level, continuous identification duration, target centroid coordinates and pixel size.

[0082] S2. Construct a confidence weight function with time as the independent variable.

[0083] S3. Calculate the overall confidence level of the target by using the confidence weight function for all identified targets.

[0084] S4. Establish a conversion function from numerical parameters of the recognition result to image display parameters. Take the target comprehensive confidence level and continuous recognition time as the function input, and select the image display parameters such as the color parameter value and line width value of the target prompt box as the function output.

[0085] S5. The ground control and display device overlays all identified target information onto the image transmitted from the missile based on the calculation results of the conversion function, so as to help the operator identify and confirm the identified targets.

[0086] Example 2

[0087] This embodiment provides a target ground-assisted identification result display method applicable to image terminal guidance munitions. In this method, the ground operation and display device receives the target ground-assisted identification result from the image transmitted from the missile, constructs a weight function and calculates the target comprehensive confidence level, and converts the numerical parameters of the identification result into image display parameters that are easier to observe and understand, so as to help the operator identify and confirm the target.

[0088] The target comprehensive confidence calculation method involves constructing a weight function related to the target duration, calculating the weighted average of the recognition confidence within the target duration, and obtaining the target comprehensive confidence.

[0089] The method for converting the numerical parameters of the recognition result into image display parameters involves establishing a conversion function from the numerical parameters of the recognition result to the image display parameters, taking the target comprehensive confidence level and continuous recognition duration as function inputs; and taking the target prompt box RGB parameter values, line width values, and other image display parameters as function outputs. The image display parameters include, but are not limited to, the above forms.

[0090] Example 3

[0091] This embodiment provides a ground display method for target identification results of image-guided munitions, which specifically includes the following steps:

[0092] S1. The ground control and display device receives the target ground-aided identification results of the missile-borne image. The identification results include information such as the number of all identified targets in the image, identification confidence level, continuous identification duration, target centroid coordinates and pixel size.

[0093] Wherein, the confidence level of target i can be expressed as W i The confidence threshold is set to 0.3, representing the degree of matching of the identification results. Therefore, the identification confidence level satisfies W. i The target i is ∈ [0.3, 1]; the duration of continuous recognition of target i can be expressed as T. i The centroid coordinates of target i can be represented by the horizontal and vertical pixel values ​​of its distance from the top left corner of the image as (Nx i ,Ny i The pixel size of target i is represented by its horizontal and vertical pixel values ​​as (Lx i ,Ly i ).

[0094] S2. Construct a confidence weight function with time as the independent variable.

[0095] The duration of the determination period can be denoted as the weighted time T. d Select the one that generally satisfies T d The weighting function is defined in the range [0.5, 3] with time t (t∈[0,T) as the basis. d With ]) as the independent variable, the function can be expressed in the form of:

[0096]

[0097] In equation (1), Q(t) is the weight value corresponding to time t, and the physical meaning of t is the time length from the current moment.

[0098] S3. Calculate the overall confidence level of all identified targets using the confidence weighting function to characterize their confidence level within the judgment period. Specifically, follow these steps:

[0099] S31, based on weighted time T d The number of recognitions N within the range generates a set of discrete time series values ​​T. w (k), the value of which is given in expression (2):

[0100] T w (k)=k·T s k = 1, 2, ..., N (2)

[0101] Among them, T s The identification cycle for ground-assisted target identification is typically tens of milliseconds, T w (k) represents the recognition time corresponding to the kth frame previously.

[0102] S32. Generate a set of data corresponding to the discrete time series T. w (k) Target i identification confidence sequence W i (k) is used to represent the weighted time T. d The confidence level of identification within the range.

[0103] Among them, W i (k) represents the identification confidence value of the k-th frame of the received target i, where k = 1, 2, ..., N.

[0104] When the continuous identification time of target i is T i ≥T d Time, that is, the number of times target i is identified N within the time period. i =N, directly take the corresponding N frames within the time range to identify the confidence level to form a sequence W. i .

[0105] When the continuous identification time of target i is T i <T d Time, that is, the number of times target i is identified N within the time period. i <N, take the first N i The recognition confidence scores corresponding to the frames form a sequence W. i (k)Previous N i For the first item, set the value of all other items in the sequence to the lower threshold of 0.3, that is:

[0106] W i (j)=0.3 j=N i N i +1,...,N (3)

[0107] S33. Calculate the recognition confidence sequence W for target i. i The weighted average of these values ​​is the overall confidence level ω of the target. i The calculation expression is shown in equation (4):

[0108]

[0109] S4. Establish a conversion function from numerical parameters of the recognition results to image display parameters, and convert the target comprehensive confidence ω i Continuous recognition duration T i As input to the function, select the color parameter value Y of the target tooltip. i Line width value Z i The image display parameters are output as a function, and the conversion function expression is as shown in equation (5):

[0110]

[0111] S41. In this embodiment, the color parameter Y of the target prompt box i Select to display RGB values, denoted as Y. i =(R i G i B i ), and set the function to the target comprehensive confidence level ω. i Single input, confidence ω i The larger the value, the more vibrant and eye-catching the displayed color will be. The function form is as shown in expression (6):

[0112]

[0113] S42. In this embodiment, the line width value Z is set. i The function is the continuous recognition duration T. i Single input, duration T i The larger the value, the larger the line width value. The function form is as shown in expression (7):

[0114]

[0115] In equation (7), T min This is the lower limit of the time threshold, typically set to 0.5 seconds; T max This is the upper limit of the time threshold, generally related to the weighted time T. d Take the same value.

[0116] S5. The ground control and display device overlays all identified target information onto the image transmitted from the missile based on the calculation results of the conversion function, so as to help the operator identify and confirm the identified targets.

[0117] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A ground-based method for displaying target recognition results of image-guided munitions, characterized in that, The method includes the following steps: Step S1: The ground control and display device receives the target ground-assisted identification result of the missile-borne image. The target ground-assisted identification result includes the number of all identified targets in the image, identification confidence level, continuous identification duration, target centroid coordinates and pixel size information. Step S2: Construct a confidence weight function with time as the independent variable; Step S3: Calculate the comprehensive confidence level of the target by using the confidence weight function for all identified targets. Step S4: Establish a conversion function from numerical parameters of the recognition result to image display parameters. Take the target comprehensive confidence and continuous recognition time as the function input, and select image display parameters including the color parameter value of the target prompt box and the line width value as the function output. Step S5: The ground control and display device overlays all identified target information onto the image screen transmitted from the missile based on the calculation results of the conversion function, so as to help the operator identify and confirm the identified targets. In step S1, the target The confidence level of identification is expressed as This characterizes the degree of matching in this identification result. With a lower limit of confidence threshold set at 0.3, the identification confidence level meets the following criteria: scope; In step S1, the target The duration of continuous identification is expressed as ;Target The centroid coordinates are expressed in horizontal and vertical pixel values ​​from the top left corner of the image. ;Target The pixel size is expressed as the horizontal and vertical pixel values ​​it occupies. ; In step S2, a confidence weight function with time as the independent variable is constructed. The duration of the determination period is recorded as the weighted time. Select the one that satisfies Range, weighting function in time , The independent variable is expressed in functional form as: （1） In equation (1), For time The corresponding weight value, The physical meaning of is the time elapsed since the current moment; In step S3, a confidence weighting function is used to calculate the comprehensive confidence score of all identified targets to characterize the confidence level of the targets within the judgment period; specifically, this is performed as follows: Step S31: Based on weighted time Number of recognitions within the range Generate a set of discrete time series values The value can be found in expression (2): （2） in, The identification cycle for ground-assisted target identification. Representing the previous The frame is identified based on the corresponding time. Step S32: Generate a set of data corresponding to the discrete time series. goal Identify confidence sequences Used to represent weighted time The confidence level of identification within the range; in, To receive the target No. The confidence score for frame identification. ; When the target Continuous recognition duration Time, that is, determining the target within a time period. Number of recognitions Directly take the corresponding time range Frame recognition confidence formation sequence ; When the target Continuous recognition duration Time, that is, determining the target within a time period. Number of recognitions Take the front The recognition confidence sequence corresponding to the frame forward For the first item, set the value of all other items in the sequence to the lower threshold of 0.3, that is: （3） Step S33: Calculate the target Identification confidence sequence The weighted average is the target overall confidence level. The calculation expression is shown in equation (4): （4）。 2. The ground display method for target recognition results of image-guided munitions as described in claim 1, characterized in that, In step S31 It takes tens of milliseconds.

3. The ground display method for target recognition results of image-guided munitions as described in claim 2, characterized in that, In step S4, a conversion function is established to transform the numerical parameters of the recognition result into image display parameters, and the target comprehensive confidence level is calculated. Continuous identification duration As input to the function, select the color parameter value of the target tooltip. Line width value The image display parameters are used as the function output, and the conversion function expression is as shown in equation (5): （5）。 4. The ground display method for target recognition results of image-guided munitions as described in claim 3, characterized in that, Step S4 includes: Step S41, Color parameters of the target tooltip Select to display RGB values, denoted as And set the function to the target comprehensive confidence level. Single input, confidence level The larger the value, the more vibrant and eye-catching the displayed color becomes. The function form is as shown in expression (6): （6） Step S42: Set the line width value The function is the duration of continuous recognition. Single input, duration The larger the value, the larger the line width value. The function form is as shown in expression (7): （7） In equation (7), The lower limit of the time threshold is set to 0.5 seconds; The upper limit of the time threshold, and the weighted time. Take the same value.

5. The ground display method for target recognition results of image-guided munitions as described in claim 3, characterized in that, The method described belongs to the field of weapons and ammunition.

6. The ground display method for target recognition results of image-guided munitions as described in claim 3, characterized in that, The method obtains numerical parameters that characterize the "similarity" of targets by constructing a confidence weight function, and converts the recognition results into image display parameters that are easier to observe and understand. The results are displayed through prompt boxes on the ground control display device, which can effectively show the "similarity" information of the recognition results and help operators to identify multiple targets in sequence according to the prompts, thereby improving the success rate of target recognition.

Images