Target detection and classification system and method based on data fusion

Abstract

The invention discloses a target detection and classification system and method based on data fusion, and relates to the technical field of target recognition and detection, and the system comprises a target detection and classification module, a to-be-worked area planning module, a flight data analysis module, and a target detection and recognition optimization module. The target detection and classification module is used for carrying out image recognition and voiceprint recognition on a target and judging the type of the target, the to-be-worked area planning module is used for carrying out space segmentation on a data acquisition effective area, historical image recognition and voiceprint recognition data of the target in the segmented space are analyzed, and an optimal fusion recognition area is planned. The probability that the target flies into the optimal fusion recognition area is analyzed through a flight data analysis module, a target detection reference model is established through a target detection recognition optimization module, the probability that the current target flies to the optimal fusion recognition area is predicted, and a detection recognition mode for the current target is planned. And the collection of invalid data in the target detection and identification process is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition, and specifically to an object detection and classification system and method based on data fusion. Background Art

[0002] Through object detection technology, it is possible to monitor and identify the activities of flying objects around some areas in real time. To reduce the missed recognition rate and improve the recognition accuracy, a combination of image recognition and voiceprint recognition can be used for object recognition and detection. The image recognition technology is: using a camera to capture the images and video footage of the object, and comparing the captured information with the object features in the database to determine the type of the object. The voiceprint recognition technology is: recording the calls of the object through a voice recorder, and comparing the collected calls of the object with the voice features in the known voiceprint database to determine the type of the object; However, both image recognition and voiceprint recognition have their respective optimal recognition areas or ranges. The two optimal recognition areas may have an overlapping area. Taking images and collecting sounds after the object appears in the overlapping area can improve the accuracy of subsequent object type recognition. In the prior art, when performing image recognition and voiceprint recognition on an object, the working matching degree between data collection devices is not high. Generally, when the object appears within the recognizable range of the device, the device starts to collect object data and then performs its respective object recognition work. At this time, there may be a large amount of data in the collected object data that leads to low recognition accuracy, which is not conducive to subsequent recognition work, and the object recognition efficiency and accuracy cannot be improved. Summary of the Invention

[0003] The purpose of the present invention is to provide an object detection and classification system and method based on data fusion to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: An object detection and classification system based on data fusion, the system includes an object detection and classification module, a to-be-worked area planning module, a flight data analysis module, and an object detection and recognition optimization module; Through the object detection and classification module, image recognition and voiceprint recognition are performed on the object to determine the type to which the object belongs; Through the to-be-worked area planning module, the effective image acquisition area and the effective voiceprint acquisition area are spatially segmented, and the historical image recognition and voiceprint recognition data of the objects in the segmented space are analyzed, and the optimal fusion recognition area is planned based on the analysis results; Through the flight data analysis module, the flight trajectory data of historical objects is obtained, a reference trajectory is selected, and the probability that the object flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different is analyzed; The target detection reference model is established through the target detection and recognition optimization module to predict the probability that the current target will fly to the optimal fusion recognition area, and the detection and recognition method for the current target is planned based on the prediction result.

[0005] Preferably, the target detection and classification module includes an image data acquisition unit, a sound data collection unit, and a target detection and classification unit; The image data acquisition unit uses an image capture device to acquire target images and transmits the target image data to the target detection and classification unit; The sound data collection unit uses a sound recorder to record the calls of the target and transmits the collected call data of the target to the target detection and classification unit; The target detection and classification unit compares the target image data with the target features in the database to determine the type of the target; compares the call of the target with the sound features in the known voiceprint database to determine the type of the target.

[0006] Preferably, the working area planning module includes an identification area segmentation unit, a sub-area information statistics unit, and an optimal fusion recognition area analysis unit; The identification area segmentation unit divides a number of spherical spaces centered on the location of the image capture device, and takes the largest spherical space as the first spherical space. The first spherical space is the maximum data acquisition range of the image capture device, that is, the image capture device cannot capture images outside the first spherical space; divides a number of spherical spaces centered on the location of the sound recorder, and takes the largest spherical space as the second spherical space. The second spherical space is the maximum data collection range of the sound recorder, that is, the sound recorder cannot collect sounds outside the second spherical space; The sub-area information statistics unit counts the number of historical target images captured by the image capture device when historical targets appear in a number of spherical spaces and the number of times the sound recorder collects the calls of the target, counts the number of times the type of the target is correctly determined after feature comparison of the target based on the historical target images, and counts the number of times the type of the target is correctly determined after feature comparison of the calls of the target collected multiple times; The optimal fusion recognition area analysis unit selects the optimal image acquisition area of the image capture device and the optimal sound collection area of the sound recorder based on the statistical results, and takes the overlapping area of the optimal image acquisition area and the optimal sound collection area as the optimal fusion recognition area.

[0007] Preferably, the flight data analysis module includes a historical flight data acquisition unit, a coincidence degree analysis unit, and a position and area matching probability analysis unit; Obtain the total number of historical targets that have flown into the first spherical space and the second spherical space in the past, as well as the number of historical targets that have flown into the optimal fusion recognition area in the past, and obtain the flight trajectories of the historical targets; Extract the flight trajectory of a randomly selected historical target that has flown into the optimal fusion recognition area through the coincidence degree analysis unit as the reference trajectory, and analyze the coincidence degree between the flight trajectories of the remaining targets and the reference trajectory. The remaining targets refer to the targets that have flown into the optimal fusion recognition area except for the extracted target; Through the position and area matching probability analysis unit, count the proportion of the number of targets in different spaces when the coincidence degree with the reference trajectory is different, and analyze the probability that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different.

[0008] Preferably, the target detection and recognition optimization module includes a target detection reference model establishment unit and a detection method planning unit; Establish a target detection reference model with the coincidence degree and probability of the flight trajectories between targets as training samples through the target detection reference model establishment unit; Obtain the coincidence degree between the generated flight trajectory of the current target and the reference trajectory through the detection method planning unit, input it into the target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result.

[0009] A target detection and classification method based on data fusion includes the following steps: S100: Perform image recognition and voiceprint recognition on the target to determine the type of the target; S101: Perform spatial segmentation on the effective image acquisition area and the effective voiceprint acquisition area, analyze the historical image recognition and voiceprint recognition data of the targets in the segmented space, and plan the optimal fusion recognition area according to the analysis results; S102: Obtain the flight trajectory data of historical targets, select the reference trajectory, and analyze the probability that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different; S103: Establish a target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result.

[0010] Preferably, the S100 includes: using an image capture device to capture the target image, comparing the target image data with the target features in the database to determine the type of the target; using a voice recorder to record the calls of the target, and comparing the collected calls of the target with the voice features in the known voiceprint database to determine the type of the target.

[0011] Preferably, S101 includes: dividing n spherical spaces with the position of the image capturing device as the center of the sphere. The diameters of the n spherical spaces are {R1, R2,... R n}, where R i > R i-1 and the difference between adjacent two diameters is a fixed value. i refers to the i-th spherical space, i = 2, 3,... n, that is, the difference between the diameters of any two adjacent terms is equal. The spherical space with the largest diameter among the n spaces is taken as the first spherical space, and the first spherical space is the maximum data acquisition range of the image capturing device. m spherical spaces with the position of the sound recorder as the center of the sphere are divided in the same way, and the spherical space with the largest diameter among the m spaces is taken as the second spherical space, and the second spherical space is the maximum data acquisition range of the sound recorder. When it is statistically found that a historical target appears in a random one of the n spaces, the number of images of the corresponding historical target captured by the image capturing device is U. After identifying K of the U images, it can be correctly judged which category the corresponding historical target belongs to. Calculate the recognition accuracy rate w j , w j = K / U. The set of recognition accuracy rates for capturing and identifying images of targets appearing in the n spherical spaces is w = {w1, w2,... w j ,... w n}. The spherical space with the highest recognition accuracy rate is selected as the best image acquisition area of the image capturing device. The spherical space with the highest recognition accuracy rate is selected as the best sound collection area of the sound recorder in the same way. If there is an overlapping area between the best image acquisition area and the best sound collection area, the overlapping area of the best image acquisition area and the best sound collection area is taken as the best fusion recognition area; if there is no overlapping area, no best fusion recognition area is divided, and the target is recognized by the method of separate recognition and judgment, that is, when the current target appears in the first spherical space, the target image is captured, and when the current target appears in the second spherical space, the target call is collected, and image recognition and voiceprint recognition are respectively performed, and their respective recognition results are transmitted to the monitoring center; Based on the purpose of the detection method for planning image and sound data acquisition, considering that the target can be correctly recognized when it does not appear at all positions within the effective acquisition range of the data acquisition device, the acquisition space is divided into several spherical spaces centered on the device, and the diameter difference between any two of them is equal, ensuring the fairness and rationality of setting the optimal data acquisition area in the follow-up. After the space segmentation is completed, analyze the recognition accuracy rate of collecting data and recognizing the target that appears in all spherical spaces, and set the optimal image acquisition area and the optimal sound collection area according to the recognition accuracy rate. If there is an overlapping area between the areas, it is judged that when the target appears in the overlapping area, the data acquisition and target recognition accuracy rate at two levels are higher. Therefore, the overlapping area is set as the optimal fusion recognition area.

[0012] Preferably, the S102 includes: obtaining that the total number of historical targets flying into the first spherical space or the second spherical space in the past is A, obtaining the number of historical targets flying into the optimal fusion recognition area in the past, obtaining the flight trajectories of the historical targets, extracting the flight trajectory of a randomly selected historical target flying into the optimal fusion recognition area as the reference trajectory, the length of the reference trajectory is L1, setting k discrete points on the reference trajectory, obtaining the discrete point coordinate set, traversing the flight trajectories of the remaining targets, counting the number of coordinates in the discrete point coordinate set existing in the flight trajectory of any other target as g, and calculating the coincidence degree W between the flight trajectory of any other target and the reference trajectory e : W e =[g / min(D e ,L1)]*100%; Among them, D e represents the length of the flight trajectory of any other target, and the obtained coincidence degree set is W={W1,W2,...W e ,...W k}, k is the number of the remaining targets, it is counted that among the A historical targets, a total of B1 targets' flight trajectories have a coincidence degree of W1 with the reference trajectory, and the probability P1 that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W1 is obtained, P1 = B1 / A, and the obtained probability set is P={P1,P2,...P k}, P k represents the probability that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W k .

[0013] Preferably, the S103 includes: forming the training samples {(W1,P1), (W2,P2),...(W k ,P kAfter fitting the training samples, a target detection reference model is established: y = φ1 * x + φ2, where φ1 and φ2 represent the fitting coefficients, and the overlap degree between the generated flight trajectory of the current target and the reference trajectory is obtained as W. ’ Let x = W. ’ The probability that the current target will fly to the optimal fusion recognition area is predicted as φ1 * W. ’ + φ2. Set the probability threshold as P. ’ If φ1 * W ’ + φ2 > P ’ It is predicted that the current target will fly to the optimal fusion recognition area, and the image and sound data acquisition work is paused. When the current target flies into the optimal fusion recognition area, the image and sound data are collected simultaneously, and the category to which the current target belongs is identified. If the image recognition result and the voiceprint recognition result are the same, the category to which the current target belongs is confirmed; otherwise, the image and sound data are collected again until the category to which the current target belongs is confirmed, and the information on the category to which the current target belongs is transmitted to the monitoring center; if φ1 * W ’ + φ2 ≤ P ’ It is predicted that the current target will not fly to the optimal fusion recognition area, the image and sound data acquisition work is not paused, the image recognition and voiceprint recognition are performed respectively, and their respective recognition results are transmitted to the monitoring center; By collecting historical data and analyzing the flight trajectories of the targets, the flight trajectory of one of the targets that appears in the optimal fusion recognition area is used as the reference trajectory, and the probability that the target will fly into the optimal fusion recognition area when the overlap degree with the reference trajectory is different is analyzed. The overlap degree and probability data are formed into training samples, and a target detection reference model is established. Based on the model, it is predicted whether the current target will fly into the optimal fusion recognition area. If so, the data acquisition is paused for a period of time, and the image and sound data are collected simultaneously after the target flies into the optimal fusion recognition area, reducing the invalid data that leads to low recognition accuracy and improving the target recognition efficiency and accuracy.

[0014] Compared with the prior art, the beneficial effects of the present invention are: The present invention divides the acquisition space into several spherical spaces centered on the device, and the diameter difference between any two of them is equal, ensuring the fairness and reasonableness of setting the optimal data acquisition area in the follow-up. After the space segmentation is completed, the recognition accuracy rate of collecting data and identifying the targets that appear in all spherical spaces is analyzed, and the optimal image acquisition area and the optimal sound collection area are set according to the recognition accuracy rate. If there is an overlapping area between the areas, it is judged that when the target appears in the overlapping area, the data acquisition and target recognition accuracy rate on two levels are higher. Therefore, the overlapping area is set as the optimal fusion recognition area; After setting the optimal fusion recognition area, by collecting historical data and analyzing the flight trajectory of the target, the flight trajectory of one of the targets that appears in the optimal fusion recognition area is used as the reference trajectory. Analyze the probability that the target will fly into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different. Combine the coincidence degree and probability data to form a training sample, and establish a target detection reference model. Use the model as a reference to predict whether the current target will fly into the optimal fusion recognition area. If so, suspend the data collection for a period of time, and then collect image and sound data simultaneously after the target flies into the optimal fusion recognition area, reducing the invalid data that leads to low recognition accuracy and improving the target recognition efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 FIG. is a schematic structural diagram of a target detection and classification system based on data fusion according to the present invention; Figure 2 FIG. is a schematic flow diagram of a target detection and classification method based on data fusion according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment 1: As Figure 1 shown, this embodiment provides a target detection and classification system based on data fusion, and the system includes: a target detection and classification module, a to-be-worked area planning module, a flight data analysis module, and a target detection and recognition optimization module; Use the target detection and classification module to perform image recognition and voiceprint recognition on the target to determine the type of the target; Use the to-be-worked area planning module to spatially divide the effective image acquisition area and the effective voiceprint acquisition area, analyze the historical image recognition and voiceprint recognition data of the targets in the divided space, and plan the optimal fusion recognition area based on the analysis results; Use the flight data analysis module to obtain the flight trajectory data of historical targets, select the reference trajectory, and analyze the probability that the target will fly into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different; Use the target detection and recognition optimization module to establish a target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target based on the prediction results.

[0018] The target detection and classification module includes an image data acquisition unit, a sound data collection unit, and a target detection and classification unit. The image data acquisition unit uses an image capture device to acquire target images and transmits the target image data to the target detection and classification unit. The sound data collection unit uses a sound recorder to record the calls of the target and transmits the collected call data of the target to the target detection and classification unit. The target detection and classification unit compares the target image data with the target features in the database to determine the category to which the target belongs. The calls of the target are compared with the sound features in the known voiceprint database to determine the category to which the target belongs.

[0019] The working area planning module to be processed includes an identification area segmentation unit, a sub-area information statistics unit, and an optimal fusion identification area analysis unit. The identification area segmentation unit divides several spherical spaces centered on the location of the image capture device, and takes the largest spherical space as the first spherical space. The first spherical space is the maximum data acquisition range of the image capture device, that is, the image capture device cannot capture images outside the first spherical space. It divides several spherical spaces centered on the location of the sound recorder, and takes the largest spherical space as the second spherical space. The second spherical space is the maximum data acquisition range of the sound recorder, that is, the sound recorder cannot collect sounds outside the second spherical space. The sub-area information statistics unit counts the number of historical target images captured by the image capture device and the number of times the sound recorder collects the calls of the target when the historical target appears in several spherical spaces, counts the number of times the category to which the target belongs is correctly determined after feature comparison of the target based on the historical target images, and counts the number of times the category to which the target belongs is correctly determined after feature comparison of the calls of the target collected several times. The optimal fusion identification area analysis unit selects the optimal image acquisition area of the image capture device and the optimal sound collection area of the sound recorder based on the statistical results, and takes the overlapping area of the optimal image acquisition area and the optimal sound collection area as the optimal fusion identification area.

[0020] The flight data analysis module includes a historical flight data acquisition unit, a coincidence degree analysis unit, and a position and area matching probability analysis unit. The historical flight data acquisition unit is used to obtain the total number of historical targets that have flown into the first spherical space and the second spherical space in the past, as well as the number of historical targets that have flown into the optimal fusion recognition area in the past, and obtain the flight trajectories of the historical targets. The coincidence degree analysis unit is used to extract the flight trajectory of a randomly selected historical target that has flown into the optimal fusion recognition area as the reference trajectory, and analyze the coincidence degree between the flight trajectories of the remaining targets and the reference trajectory. The remaining targets refer to the targets that have flown into the optimal fusion recognition area except for the selected target. The position and area matching probability analysis unit is used to count the proportion of the number of targets in different spaces when the coincidence degree with the reference trajectory is different, and analyze the probability that the target will fly into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different.

[0021] The target detection and recognition optimization module includes a target detection reference model establishment unit and a detection method planning unit. The target detection reference model establishment unit is used to establish a target detection reference model with the coincidence degree and probability of the flight trajectories between targets as training samples. The detection method planning unit is used to obtain the coincidence degree between the generated flight trajectory of the current target and the reference trajectory, input it into the target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result.

[0022] Embodiment 2: As Figure 2 shown, this embodiment provides a target detection and classification method based on data fusion, which is implemented based on the detection system in the embodiment, and specifically includes the following steps: S100: Perform image recognition and voiceprint recognition on the target to determine the type of the target: Use an image capture device to collect the target image, compare the target image data with the target features in the database, and determine the type of the target; Use a sound recorder to record the calls of the target, compare the collected calls of the target with the voiceprint features in the known voiceprint database, and determine the type of the target. S101: Perform spatial segmentation on the effective image acquisition area and the effective voiceprint acquisition area, analyze the historical image recognition and voiceprint recognition data of the targets in the segmented space, and plan the optimal fusion recognition area according to the analysis results: Divide n spherical spaces with the position of the image capture device as the center of the sphere. The diameters of the n spherical spaces are {R1, R2,... R n}, where R i >R i-1And the difference between two adjacent diameters is a fixed value. Here, i represents the i-th spherical space, where i = 2, 3,... n. That is, the difference between the diameters of any two adjacent terms is equal. The spherical space with the largest diameter among the n spaces is taken as the first spherical space, and the first spherical space is the maximum data acquisition range of the image capture device. m spherical spaces centered at the location of the sound recorder are divided in the same way. The spherical space with the largest diameter among the m spaces is taken as the second spherical space, and the second spherical space is the maximum data acquisition range of the sound recorder. When it is statistically found that a historical target appears in a random spherical space among the n spaces, the number of images of the corresponding historical target captured by the image capture device is U. After identifying K out of the U images, it is correctly determined which category the corresponding historical target belongs to. The recognition accuracy rate w for capturing and identifying images of a target appearing in a random spherical space is calculated. j , w j = K / U. The set of recognition accuracy rates for capturing and identifying images of targets appearing in the n spherical spaces is w = {w1, w2,... w j ,... w n}. The spherical space with the highest recognition accuracy rate is selected as the best image acquisition area of the image capture device. In the same way, the spherical space with the highest recognition accuracy rate is selected as the best sound collection area of the sound recorder. The screening method for the best sound collection area is the same as that for the best image acquisition area: When it is statistically found that a historical target appears in a random spherical space among the m spaces, the sound recorder has collected the calls of the corresponding historical target a total of a times. After performing voiceprint recognition on b of the collected calls, it is correctly determined which category the target belongs to. The recognition accuracy rate for capturing and identifying the sound of a target appearing in a random spherical space among the m spherical spaces is b / a. The recognition accuracy rate for capturing and identifying the sound of targets appearing in the m spherical spaces is obtained, and the spherical space with the highest recognition accuracy rate is selected as the best sound collection area of the sound recorder. If there is an overlapping area between the best image acquisition area and the best sound collection area, the overlapping area of the best image acquisition area and the best sound collection area is taken as the best fusion recognition area; if there is no overlapping area, no best fusion recognition area is divided, and the target is recognized by the method of separate recognition and judgment. That is, when the current target appears in the first spherical space, the target image is captured, and when the current target appears in the second spherical space, the target call is collected. Image recognition and voiceprint recognition are performed separately, and their respective recognition results are transmitted to the monitoring center; S102: Obtain the flight trajectory data of historical targets, select a reference trajectory, and analyze the probability that a target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different: Obtain that the total number of historical targets that flew into the first spherical space or the second spherical space in the past is A, obtain the number of historical targets that flew into the optimal fusion recognition area in the past, obtain the flight trajectories of historical targets, extract the flight trajectory of a randomly selected historical target that flew into the optimal fusion recognition area as the reference trajectory, the length of the reference trajectory is L1, set k discrete points on the reference trajectory, obtain the discrete point coordinate set, traverse the flight trajectories of the remaining targets, count the number of coordinates in the discrete point coordinate set that exist in the flight trajectory of another randomly selected target as g, and calculate the coincidence degree W between the flight trajectory of another randomly selected target and the reference trajectory. e : W e = [g / min(D e , L1)] * 100%; Among them, D e represents the length of the flight trajectory of another randomly selected target, and the obtained coincidence degree set is W = {W1, W2,... W e ,... W k}, k is the number of the remaining targets, count that among the A historical targets, a total of B1 targets have a coincidence degree of W1 with the reference trajectory, and obtain the probability P1 that a target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W1, P1 = B1 / A, and obtain the probability set as P = {P1, P2,... P k}, P k represents the probability that a target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W k ; S103: Establish a target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result: Compose the training samples {(W1, P1), (W2, P2),... (W k , P k ), and establish a target detection reference model after fitting the training samples: y = φ1 * x + φ2, where φ1 and φ2 represent the fitting coefficients, * represents the multiplication sign, obtain that the coincidence degree between the generated flight trajectory of the current target and the reference trajectory is W ’ , let x = W ’ , and predict that the probability that the current target will fly into the optimal fusion recognition area is φ1 * W ’ +φ2, set the probability threshold as P ’ , if φ1 * W ’ +φ2 > P ’, predict that the current target will fly to the optimal fusion recognition area, pause the image and sound data collection work, and resume the image and sound data collection simultaneously when the current target flies into the optimal fusion recognition area. Identify the category to which the current target belongs. If the image recognition result and the voiceprint recognition result are the same, confirm the category to which the current target belongs; otherwise, resume the image and sound data collection until the category to which the current target belongs is confirmed, and transmit the confirmed category information of the current target to the monitoring center; if φ1*W ’ +φ2≤P ’ , predict that the current target will not fly to the optimal fusion recognition area, do not pause the image and sound data collection work, perform image recognition and voiceprint recognition separately, and transmit their respective recognition results to the monitoring center respectively; For example: Taking the target as a flying bird as an example: When detecting and identifying the target, the coincidence degree between the obtained flight trajectory of the current flying bird and the reference trajectory is 0.83. After inputting 0.83 into the target detection reference model, the predicted probability that the current flying bird will fly to the optimal fusion recognition area is 0.90. Set the probability threshold to 0.85, 0.90>0.85. Predict that the current flying bird will fly to the optimal fusion recognition area, pause the image and sound data collection work, and resume the image and sound data collection simultaneously when the current flying bird flies into the optimal fusion recognition area. Identify the category to which the current flying bird belongs. If the image recognition result and the voiceprint recognition result are the same, confirm the category to which the current flying bird belongs.

[0023] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claimed claims.

Claims

1. A target detection and classification system based on data fusion, characterized in that: The system includes: a target detection and classification module, a working area planning module, a flight data analysis module, and a target detection and recognition optimization module; The target detection and classification module performs image recognition and voiceprint recognition on the target to determine the type of the target; The working area planning module spatially divides the effective image acquisition area and the effective voiceprint acquisition area, analyzes the historical image recognition and voiceprint recognition data of the targets in the divided space, and plans the best fusion recognition area based on the analysis results; The flight data analysis module obtains the flight trajectory data of historical targets, selects a reference trajectory, and analyzes the probability that the target flies into the best fusion recognition area when the coincidence degree with the reference trajectory is different; The target detection and recognition optimization module establishes a target detection reference model, predicts the probability that the current target will fly into the best fusion recognition area, and plans the detection and recognition method for the current target based on the prediction results.

2. The object detection and classification system based on data fusion according to claim 1, wherein: The target detection and classification module includes an image data acquisition unit, a sound data collection unit, and a target detection and classification unit; The image data acquisition unit uses an image capture device to capture target images and transmits the target image data to the target detection and classification unit; The sound data collection unit uses a sound recorder to record the calls of the target and transmits the collected call data of the target to the target detection and classification unit; The target detection and classification unit compares the target image data with the target features in the database to determine the type of the target; The calls of the target are compared with the voiceprint features in the known voiceprint database to determine the type of the target.

3. The object detection and classification system based on data fusion according to claim 2, characterized in that: The working area planning module includes an identification area segmentation unit, a sub-area information statistics unit, and a best fusion recognition area analysis unit; The identification area segmentation unit divides a number of spherical spaces centered on the location of the image capture device, and takes the largest spherical space as the first spherical space, which is the maximum data acquisition range of the image capture device; divides a number of spherical spaces centered on the location of the sound recorder, and takes the largest spherical space as the second spherical space, which is the maximum data acquisition range of the sound recorder; The sub-area information statistics unit counts the number of historical target images captured by the image capture device when historical targets appear in a number of spherical spaces and the number of times the sound recorder collects the calls of the target, counts the number of times the type of the target is correctly determined by comparing the features based on the historical target images, and counts the number of times the type of the target is correctly determined by comparing the features of the calls of the target collected several times; The best fusion recognition area analysis unit selects the best image acquisition area of the image capture device and the best sound collection area of the sound recorder based on the statistical results, and takes the overlapping area of the best image acquisition area and the best sound collection area as the best fusion recognition area.

4. The object detection and classification system based on data fusion according to claim 3, characterized in that: The flight data analysis module includes a historical flight data acquisition unit, a coincidence degree analysis unit, and a position and area matching probability analysis unit; Obtain the total number of historical targets that have flown into the first spherical space and the second spherical space in the past, as well as the number of historical targets that have flown into the optimal fusion recognition area in the past, and obtain the flight trajectories of the historical targets through the historical flight data acquisition unit; Retrieve the flight trajectory of a randomly selected historical target that has flown into the optimal fusion recognition area as the reference trajectory through the coincidence degree analysis unit, and analyze the coincidence degree between the flight trajectories of the remaining targets and the reference trajectory; Statistically analyze the proportion of the number of targets in different spaces when the coincidence degree with the reference trajectory is different through the position and area matching probability analysis unit, and analyze the probability that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different.

5. The object detection and classification system based on data fusion according to claim 4, characterized in that: The target detection and recognition optimization module includes a target detection reference model establishment unit and a detection method planning unit; Establish a target detection reference model with the coincidence degree and probability of the flight trajectories between targets as training samples through the target detection reference model establishment unit; Obtain the coincidence degree between the generated flight trajectory of the current target and the reference trajectory through the detection method planning unit, input it into the target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result.

6. A target detection and classification method based on data fusion, characterized in that: It includes the following steps: S100: Perform image recognition and voiceprint recognition on the target to determine the type of the target; S101: Perform spatial segmentation on the effective image acquisition area and the effective voiceprint acquisition area, analyze the historical image recognition and voiceprint recognition data of the targets in the segmented space, and plan the optimal fusion recognition area according to the analysis results; S102: Obtain the flight trajectory data of historical targets, select the reference trajectory, and analyze the probability that the target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is different; S103: Establish a target detection reference model, predict the probability that the current target will fly into the optimal fusion recognition area, and plan the detection and recognition method for the current target according to the prediction result.

7. A target detection and classification method based on data fusion according to claim 6, characterized in that: The S100 includes: using an image capture device to capture the target image, comparing the target image data with the target features in the database to determine the type of the target; using a voice recorder to record the calls of the target, and comparing the collected calls of the target with the voice features in the known voiceprint database to determine the type of the target.

8. A method for object detection and classification based on data fusion according to claim 7, characterized in that: The S101 includes: dividing n spherical spaces with the location of the image capture device as the center of the sphere. The diameters of the n spherical spaces are {R1, R2,... R n}, where R i > R i-1 and the difference between adjacent diameters is a fixed value. i refers to the i-th spherical space, i = 2, 3,... n. The spherical space with the largest diameter among the n spaces is taken as the first spherical space, and the first spherical space is the maximum data acquisition range of the image capture device. m spherical spaces with the location of the sound recorder as the center of the sphere are divided in the same way, and the spherical space with the largest diameter among the m spaces is taken as the second spherical space, and the second spherical space is the maximum data acquisition range of the sound recorder. When it is statistically found that a historical target appears in a random one of the n spaces, the number of images of the corresponding historical target captured by the image capture device is U. After identifying K of the U images, it is possible to correctly determine the category to which the corresponding historical target belongs. Calculate the recognition accuracy rate w j of collecting and identifying images of the target appearing in a random spherical space. w j = K / U. The set of recognition accuracy rates of collecting and identifying images of the target appearing in the n spherical spaces is w = {w1, w2,... w j ,... w n}. The spherical space with the highest recognition accuracy rate is selected as the best image acquisition area of the image capture device. The spherical space with the highest recognition accuracy rate is selected as the best sound collection area of the sound recorder in the same way. If there is an overlapping area between the best image acquisition area and the best sound collection area, the overlapping area of the best image acquisition area and the best sound collection area is taken as the best fusion recognition area; if there is no overlapping area, no best fusion recognition area is divided.

9. The object detection and classification method based on data fusion according to claim 8, wherein: The S102 includes: obtaining that the total number of historical targets flying into the first spherical space or the second spherical space in the past is A, obtaining the number of historical targets flying into the optimal fusion recognition area in the past, obtaining the flight trajectories of the historical targets, extracting the flight trajectory of a randomly selected historical target flying into the optimal fusion recognition area as the reference trajectory, the length of the reference trajectory is L1, setting k discrete points on the reference trajectory, obtaining the discrete point coordinate set, traversing the flight trajectories of the remaining targets, counting that the number of coordinates existing in the discrete point coordinate set in the flight trajectory of another randomly selected target is g, and calculating the coincidence degree W between the flight trajectory of another randomly selected target and the reference trajectory e : W e =[g / min(D e ,L1)]*100%; Among them, D e represents the flight trajectory length of any one of the remaining targets, and the coincidence degree set is W = {W1, W2,... W e ,... W k}, where k is the number of the remaining targets. It is counted that among A historical targets, there are B1 targets whose flight trajectories coincide with the reference trajectory with a coincidence degree of W1. The probability P1 that a target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W1 is P1 = B1 / A. The probability set is P = {P1, P2,... P k}, and P k represents the probability that a target flies into the optimal fusion recognition area when the coincidence degree with the reference trajectory is W k .

10. A target detection and classification method based on data fusion according to claim 9, characterized in that: The S103 includes: forming training samples {(W1, P1), (W2, P2),... (W k , P k ), fitting the training samples to establish a target detection reference model: y = φ1 * x + φ2, where φ1 and φ2 represent fitting coefficients, obtaining the overlap degree W ’ between the generated flight trajectory and the reference trajectory of the current target, letting x = W ’ , predicting the probability that the current target will fly to the optimal fusion recognition area as φ1 * W ’ + φ2, setting the probability threshold as P ’ , if φ1 * W ’ + φ2 > P ’ , predicting that the current target will fly to the optimal fusion recognition area, pausing the acquisition of image and sound data, and resuming the acquisition of image and sound data simultaneously when the current target flies into the optimal fusion recognition area, identifying the category to which the current target belongs. If the image recognition result and the voiceprint recognition result are the same, then confirm the category to which the current target belongs; otherwise, re - acquire image and sound data until the category to which the current target belongs is confirmed, and transmit the confirmed category information of the current target to the monitoring center; if φ1 * W ’ + φ2 ≤ P ’ , predicting that the current target will not fly to the optimal fusion recognition area, not pausing the acquisition of image and sound data, respectively performing image recognition and voiceprint recognition, and transmitting their respective recognition results to the monitoring center.

Images