An adaptive clustering method and millimeter wave detection machine in target detection process

Abstract

The application relates to an adaptive clustering method in a target detection process and a millimeter wave detection machine. The method comprises the following steps: acquiring a data point group in a coverage range, the data point group being point cloud data generated in a unit time length; performing primary division on the data point group to obtain at least one feature data point group; checking the speed consistency of the feature data point group; when there are multiple speeds, performing splitting processing on the feature data point group, and obtaining multiple sub-feature data point groups after the splitting processing of the feature data point group; assigning a category label to the sub-feature data point group and performing contour feature checking according to the category label; and performing secondary division on the sub-feature data point group according to the contour feature checking result. The adaptive clustering method in the target detection process and the millimeter wave detection machine disclosed by the application perform multiple divisions on the data point group based on speed and contour features, and are used for identifying close-range targets or overlapping targets.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an adaptive clustering method in a target detection process and a millimeter wave detection machine. BACKGROUND

[0002] The target detection principle of the millimeter wave detection machine is to emit millimeter waves of a specific frequency band, receive the echo reflected after encountering a target, and accurately determine the position, speed and angle of the target by analyzing the difference between the echo and the emitted wave.

[0003] In the target detection process, a large number of candidate "points" or "areas" need to be classified. The purpose of classification is to determine how many targets exist in the detection area, and then to continuously track these targets. Commonly used clustering methods include density-based clustering and distance-based clustering. These methods are based on fixed parameters and have certain difficulties in processing close-range targets and overlapping targets.

[0004] In addition, there are ways to classify close-range targets and overlapping targets by increasing the dimension, but this will cause data sparsity and distance failure, and the computational complexity will increase exponentially, which will cause the result output to lag seriously in real-time processing scenarios. SUMMARY

[0005] The present application provides an adaptive clustering method in a target detection process and a millimeter wave detection machine, which divides data point groups multiple times based on speed and contour features to identify close-range targets or overlapping targets.

[0006] The above-mentioned purpose of the present application is achieved by the following technical scheme:

[0007] In a first aspect, the present application provides an adaptive clustering method in a target detection process, comprising:

[0008] Obtaining a data point group in a coverage range, the data point group being point cloud data generated in a unit time length;

[0009] Dividing the data point group once to obtain at least one feature data point group;

[0010] Checking the speed consistency of the feature data point group, and when there are multiple speeds, splitting the feature data point group, the feature data point group being split to obtain multiple sub-feature data point groups;

[0011] Assigning a class label to the sub-feature data point group and performing contour feature checking according to the class label;

[0012] According to the contour feature checking result, the sub-feature data point group is divided twice.

[0013] In a possible implementation manner of the first aspect, the one-time division on the data point group comprises:

[0014] calculating a frequency change direction of each data point in the data point group;

[0015] performing one-time screening on the data point group according to the frequency change direction, and removing invalid points in the data point group, the invalid points not having the frequency change direction;

[0016] constructing an analysis grid using the data points, each grid in the analysis grid having the same number of edges;

[0017] calculating a change trend of each grid on a time sequence, the change trend comprising an area change trend and a shape change trend;

[0018] dividing the data point group according to the change trend to obtain at least one characteristic data point group;

[0019] wherein the change trend of the grid associated with the characteristic data point group is the same.

[0020] In a possible implementation manner of the first aspect, after the data point group is divided according to the change trend to obtain at least one characteristic data point group, the method further comprises:

[0021] determining a shape of the characteristic data point group to obtain a contour feature;

[0022] checking the data point group using the contour feature to determine whether there is a missing area in the data point group;

[0023] when there is the missing area in the data point group, marking the missing area as a to-be-assigned area.

[0024] In a possible implementation manner of the first aspect, the method further comprises:

[0025] counting a frequency of occurrence of the to-be-assigned area on the time sequence;

[0026] when the frequency of occurrence of the to-be-assigned area is greater than or equal to a set frequency, determining a positional relationship between the to-be-assigned area and the characteristic data point group and calculating a distance between the to-be-assigned area and the millimeter wave detector according to the positional relationship;

[0027] selecting an auxiliary scanning area according to the positional relationship and the distance and scanning the auxiliary scanning area;

[0028] when an area of the to-be-assigned area increases or remains unchanged, regarding the to-be-assigned area as a sub-characteristic data point group.

[0029] In a possible implementation manner of the first aspect, the checking the speed consistency of the feature data point group and performing the splitting processing on the feature data point group according to the checking result comprises:

[0030] selecting a plurality of data points in the feature data point group as representative points, the representative points being located at non-edge positions of the feature data point group;

[0031] calculating a frequency variation speed and a frequency variation direction of the representative points;

[0032] determining a splitting position according to the frequency variation speed and the frequency variation direction of the representative points;

[0033] performing secondary division on the data points at the splitting position using the splitting position.

[0034] In a possible implementation manner of the first aspect, the performing the secondary division on the data points at the splitting position using the splitting position comprises:

[0035] selecting a data point region group at the splitting position as a marker data point region group;

[0036] determining a moving direction of the marker data point region group according to other data point region groups on both sides of the marker data point region group;

[0037] moving the marker data point region group in the moving direction and recording an average frequency variation speed and an average frequency variation direction of the marker data point region group;

[0038] when the average frequency variation speed or the average frequency variation direction of the marker data point region group changes, recording a change trend of the changed average frequency variation speed or average frequency variation direction;

[0039] determining a median position point of the change trend and sequentially connecting adjacent median position points to obtain a median position reference line;

[0040] performing the secondary division on the data point group at the splitting position using the median position reference line.

[0041] In a possible implementation manner of the first aspect, when performing the secondary division on the sub-feature data point group according to the contour feature checking result, the method further comprises:

[0042] determining a pending region according to the contour feature checking result;

[0043] calculating a correlation degree of the pending region and the sub-feature data point group corresponding to the contour feature checking result;

[0044] selecting, according to the correlation degree, whether to divide the pending region into the sub-feature data point group corresponding to the contour feature checking result or to keep the original division manner.

[0045] In a second aspect, the present application provides an adaptive clustering device in a target detection process, comprising:

[0046] a data acquisition unit configured to acquire a data point group in a coverage range, the data point group being point cloud data generated in a unit time length;

[0047] a first classification unit configured to divide the data point group once to obtain at least one feature data point group;

[0048] a first checking unit configured to check the speed consistency of the feature data point group, and perform splitting processing on the feature data point group when there are multiple speeds, the feature data point group being divided into multiple sub-feature data point groups after the splitting processing;

[0049] a second checking unit configured to assign a category label to the sub-feature data point group and perform contour feature checking according to the category label;

[0050] a second classification unit configured to divide the sub-feature data point group twice according to the contour feature checking result.

[0051] In a third aspect, the present application provides a millimeter wave detection machine, comprising:

[0052] one or more memories configured to store instructions;

[0053] one or more processors configured to call and run the instructions from the memories, and perform the method as described in the first aspect and any possible implementation manner of the first aspect.

[0054] In a fourth aspect, the present application provides a computer readable storage medium, comprising:

[0055] a program, when the program is run by a processor, the method as described in the first aspect and any possible implementation manner of the first aspect is performed.

[0056] In a fifth aspect, the present application provides a computer program product, comprising program instructions, when the program instructions are run by a computing device, the method as described in the first aspect and any possible implementation manner of the first aspect is performed.

[0057] In a sixth aspect, the present application provides a chip system, which comprises a processor configured to implement the functions involved in the above aspects, for example, generating, receiving, sending, or processing the data and / or information involved in the above method.

[0058] The chip system can be composed of a chip, or can include a chip and other discrete devices.

[0059] In a possible design, the chip system further includes a memory configured to store necessary program instructions and data. The processor and the memory can be decoupled and arranged on different devices and connected through wired or wireless means, or the processor and the memory can be coupled on the same device. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is a schematic block diagram of a step flow of an adaptive clustering method in a target detection process provided by the present application.

[0061] Figure 2 is a schematic diagram of a principle of detecting a target provided by the present application.

[0062] Figure 3 is a schematic diagram of an area change trend provided by the present application.

[0063] Figure 4 is a schematic diagram of a shape change trend provided by the present application.

[0064] Figure 5 is a schematic diagram of a missing area provided by the present application.

[0065] Figure 6 is a schematic diagram of a principle of calculating a distance between a to-be-allocated area and a millimeter wave detection machine provided by the present application.

[0066] Figure 7 is a schematic diagram of selecting a representative point in a feature data point group provided by the present application.

[0067] Figure 8 is a schematic diagram of a process of determining a splitting processing position by using a local-global processing manner provided by the present application.

[0068] Figure 9 is a schematic diagram of a principle of obtaining a median position point provided by the present application. DETAILED DESCRIPTION

[0069] The technical solutions in the present application are further described in detail below with reference to the drawings.

[0070] The present application discloses an adaptive clustering method in a target detection process, please refer to Figure 1 In some examples, the adaptive clustering method in a target detection process provided by the present application includes the following steps:

[0071] S101, obtaining a data point group in a coverage range, the data point group being point cloud data generated in a unit time length;

[0072] S102, performing a division on the data point group to obtain at least one feature data point group;

[0073] S103, check the speed consistency of the feature data point group, when there are multiple speeds, split the feature data point group, and obtain multiple sub-feature data point groups after splitting the feature data point group;

[0074] S104, assign a category label to the sub-feature data point group and perform contour feature checking according to the category label;

[0075] S105, according to the contour feature checking result, the sub-feature data point group is divided again.

[0076] In step S101, first, the feedback data in the coverage range is obtained, and the feedback data here is based on the feedback wave formed by the millimeter wave detector, please refer to Figure 2 , the specific process of generating data point group by detection wave and feedback wave is as follows:

[0077] Transmit: the millimeter wave detector transmits a millimeter wave signal (transmission wave) to a certain direction (φ, θ);

[0078] Propagation and reflection: the millimeter wave signal propagates in space and reflects after encountering an object;

[0079] Receive and measure distance: the receiver receives the reflected millimeter wave signal (reception wave), and uses the time of flight (ToF) method to accurately calculate the time difference Δt between transmission and reception, so as to obtain the distance d;

[0080] Record point: the system records the polar coordinates (φ, θ, d) of the current point;

[0081] Scan: quickly change the pointing direction (φ, θ) of the laser beam and repeat the above steps;

[0082] Generate data point group: in a very short time, measure millions of directions, and obtain a data point group (three-dimensional point cloud) composed of (x, y, z) coordinates, which can describe the surrounding environment.

[0083] A data point group corresponds to a time point (generally a very short time period), and all point cloud data generated at the time point forms a data point group.

[0084] After obtaining the data point group, step S102 is executed, in which the data point group is divided once to obtain at least one feature data point group, and the division method used at this time is generally distance-based division or density-based division, which is a relatively fuzzy division method.

[0085] In step S103, the speed consistency of the feature data point group is checked, when there are multiple speeds, the feature data point group is split, and after the feature data point group is split, multiple sub-feature data point groups are obtained, and the speed of the data points in each sub-feature data point group has consistency.

[0086] Then in step S104, a class label is assigned to the sub-feature data point group and a contour feature check is performed according to the class label. This step assigns a class label according to the shape of the sub-feature data point group, and then uses the class label to determine the corresponding contour, and then uses the contour feature to further divide the sub-feature data point group, that is, the content in step S105.

[0087] In the above manner, the contour feature is combined for secondary division under the preliminary division based on speed. The main purpose of the secondary division is to determine some ambiguous boundaries of the speed division. The main reasons for the ambiguous boundaries are processing errors and ambiguous regions in the boundary region. For example, at the junction of two tracking objects, if the moving speeds of the two objects are inconsistent, but the positioning of the spatial points deviates, an ambiguous boundary will appear.

[0088] In some examples, the data point group is divided once, and the application also provides other manners:

[0089] S201, calculating the frequency change direction of each data point in the data point group;

[0090] S202, performing a first screening on the data point group according to the frequency change direction, and removing invalid points in the data point group, the invalid points not having the frequency change direction;

[0091] S203, constructing an analysis grid using the data points, and the number of edges of each grid in the analysis grid being the same;

[0092] S204, calculating the change trend of each grid in the time sequence, and the change trend including the area change trend and the shape change trend;

[0093] S205, dividing the data point group according to the change trend to obtain at least one feature data point group;

[0094] The change trend of the grid associated with the feature data point group is the same.

[0095] The contents in steps S201 to S205 first need to calculate the frequency change direction of each data point in the data point group, and then perform a first screening on the data point group according to the frequency change direction. The purpose of this part is to remove invalid points in the data point group, and the invalid points here do not have the frequency change direction, that is, they are in a fixed state.

[0096] These invalid points in the fixed state are generated based on the environment, not the tracking object involved in the present application.

[0097] In the first screening of the data point group according to the frequency change direction, a suitable processing method is as follows:

[0098] Transmit a continuous wave with a linearly changing frequency over time;

[0099] At any time t, the received reflected wave is the wave transmitted at a certain time t-Δt before, so its instantaneous frequency has a difference Δf from the instantaneous frequency of the current transmission.

[0100] In this way, the continuous wave has a moving direction, and when it produces a difference Δf, the moving direction of the continuous wave is the frequency change direction of the data point.

[0101] For invalid points, because they do not have a moving state, when detected, multiple detections can be performed in a region, and the detection results (two data point groups) at this time have high similarity, and even the same result can be obtained, so direct judgment can be performed. However, for the tracking target in the moving state, it is difficult to perform point-to-point correspondence at this time.

[0102] In the present application, the solution to this problem is as follows:

[0103] First, use the data points to construct an analysis grid, and the number of edges of each grid in the analysis grid is the same, and the edge lines of each grid are required to have no intersection. Then, calculate the change trend of each grid in the time sequence. There are two change trends, which are area change trend and shape change trend. The area change trend is mainly realized by distance, and the area increases when the distance is short, as shown in Figure 3 , and the area decreases when the distance is far; the shape change trend is mainly realized by the change of the spatial pose, as shown in Figure 4 , when the spatial pose of the tracking object changes, the shape of the corresponding analysis grid will change.

[0104] In step S205, the data point group is divided according to the change trend, and at least one characteristic data point group is obtained. At this time, the change trend of the grid associated with the characteristic data point group is the same.

[0105] After the data point group is divided according to the change trend and at least one characteristic data point group is obtained, the following steps are added:

[0106] Determine the shape of the characteristic data point group to obtain the contour feature;

[0107] Use the contour feature to check the data point group to determine whether there is a missing area in the data point group;

[0108] When there is a missing area in the data point group, the missing area is marked as a to-be-assigned area.

[0109] The purpose of these steps is to determine the shape of the tracked object by means of the shape of the feature data point group, at this time a contour feature is obtained, which is obtained based on previously collected data. For example, for an aircraft, the type of aircraft needs to be determined first, and then the shape of the feature data point group is compared with the known contour feature. When the comparison is successful, the data point group is checked using the contour feature to determine whether there is a missing area in the data point group. As shown in Figure 5 When there is a missing area in the data point group, the missing area is marked as a to-be-assigned area, and when there is no missing area, this step is performed.

[0110] The feature data point group is a part of the data point group. In the foregoing steps, the data point group should be completely divided by the feature data point group. If it is not completely divided, the remaining part is classified into the to-be-assigned area (missing area).

[0111] Further, for the to-be-assigned area, the following processing is used:

[0112] The occurrence frequency of the to-be-assigned area is counted in time series;

[0113] When the occurrence frequency of the to-be-assigned area is greater than or equal to a set frequency, the positional relationship between the to-be-assigned area and the feature data point group is determined, and the distance between the to-be-assigned area and the millimeter wave detector is calculated according to the positional relationship;

[0114] An auxiliary scanning area is selected according to the positional relationship and the distance, and the auxiliary scanning area is scanned;

[0115] When the area of the to-be-assigned area increases or remains unchanged, the to-be-assigned area is taken as a sub-feature data point group.

[0116] Here, the occurrence frequency of the to-be-assigned area is used to select a suitable processing method. When the occurrence frequency of the to-be-assigned area is less than the set frequency, the main consideration is error at this time, and the processing is performed in the ignoring mode.

[0117] If the occurrence frequency of the to-be-assigned area is greater than or equal to the set frequency, the positional relationship between the to-be-assigned area and the feature data point group needs to be determined, and the distance between the to-be-assigned area and the millimeter wave detector is calculated according to the positional relationship. Then, an auxiliary scanning area is selected according to the positional relationship and the distance, and the auxiliary scanning area is scanned. At this time, if the area of the to-be-assigned area increases, the to-be-assigned area is taken as a sub-feature data point group.

[0118] Here, please refer to Figure 6The position relationship (relative position) refers to a position relationship between the to-be-assigned region and the feature data point group, and the distance refers to a distance between the to-be-assigned region and the millimeter wave detection machine. The position relationship between the feature data point group and the millimeter wave detection machine and the distance are known quantities. At this time, the distance between the to-be-assigned region and the millimeter wave detection machine can be calculated through the position relationship between the to-be-assigned region and the feature data point group.

[0119] In some examples, the speed consistency of the feature data point group is checked, and the feature data point group is split according to the checking result. The specific manner is as follows:

[0120] S301, selecting a plurality of data points in the feature data point group as representative points, the representative points being located at non-edge positions of the feature data point group;

[0121] S302, calculating a frequency change speed and a frequency change direction of the representative points;

[0122] S303, determining a split position according to the frequency change speed and the frequency change direction of the representative points;

[0123] S304, performing secondary division on data points at the split processing position using the split processing position.

[0124] In steps S301 to S305, a plurality of data points in the feature data point group are first selected as representative points, please refer to Figure 7 The representative points are located at non-edge positions of the feature data point group.

[0125] In terms of selection, the range distance is generally used for selection, the distance between two representative points is 10%-15% of the corresponding line segment distance in the feature data point group, and the representative points are uniformly distributed in the feature data point group.

[0126] Then, the frequency change speed and the frequency change direction of the representative points are calculated, and then the split position is determined according to the frequency change speed and the frequency change direction of the representative points. The specific manner is as follows: if the frequency change speed and the frequency change direction of two adjacent representative points remain consistent, it is indicated that there is no split position between the two adjacent representative points, and vice versa.

[0127] At this time, a new representative point needs to be added between the two representative points, and the above process is repeated until the split position is determined. Finally, the data points at the split processing position are divided again using the split processing position.

[0128] In some possible implementation manners, the specific manner of performing secondary division on the data points at the split processing position using the split processing position is as follows:

[0129] selecting a data point area group at the splitting processing position, denoted as a marked data point area group;

[0130] determining a moving direction of the marked data point area group according to other data point area groups on both sides of the marked data point area group;

[0131] moving the marked data point area group in the moving direction and recording an average frequency change speed and an average frequency change direction of the marked data point area group;

[0132] recording a change trend of the average frequency change speed or the average frequency change direction of the marked data point area group when the average frequency change speed or the average frequency change direction changes;

[0133] determining median position points of the change trend and sequentially connecting adjacent median position points to obtain a median position reference line;

[0134] using the median position reference line to perform secondary division on the data point group at the splitting processing position.

[0135] Please refer to Figure 8 , which adopts a local-global processing mode, aiming to quickly determine the splitting position. The specific principle is to determine the splitting position through changes in the average frequency change speed and the average frequency change direction, because the average frequency change speed or the average frequency change direction in a local moving data point area group can be represented by a curve.

[0136] The average frequency change speed refers to an average value of frequency change speeds of data points in a local moving data point area group, which is represented by a numerical value.

[0137] The average frequency change direction refers to an average value of frequency change directions of data points in a local moving data point area group, which is represented by an angle.

[0138] Please refer to Figure 9 , the horizontal coordinate of the curve is Figure 8 the position of the dashed-dotted rectangular frame, and the vertical coordinate is the cumulative value of the calculated difference between the average frequency change speed and the average frequency change direction of two adjacent dashed-dotted rectangular frames.

[0139] At this time, the median position points of the change trend can be selected as reference positions of the splitting processing position, and then these reference positions can be sequentially connected together to obtain a median position reference line. Then, the median position reference line is used to perform secondary division on the data point group at the splitting processing position.

[0140] In some examples, when the sub-feature data point group is further divided according to the contour feature inspection result, the following steps are added:

[0141] determining the pending region according to the contour feature inspection result;

[0142] calculating the correlation degree between the pending region and the sub-feature data point group corresponding to the contour feature inspection result;

[0143] dividing the pending region into the sub-feature data point group corresponding to the contour feature inspection result or keeping the original division mode according to the correlation degree.

[0144] The purpose of using this processing mode is that there is a movement error in determining the pending region according to the contour feature inspection result. The movement error is caused by the following reasons: in the comparison process of the contour feature inspection, the contour feature is a continuous change in the spatial range. If the sampling result is too dense at this time, the calculation process will be too time-consuming. Therefore, in the processing process, an interval sampling mode is generally selected, which leads to a certain amount of misjudgment probability of the pending region when the pending region is determined according to the contour feature inspection result. Therefore, the correlation degree between the pending region and the sub-feature data point group corresponding to the contour feature inspection result needs to be calculated.

[0145] The specific way of calculating the correlation degree between the pending region and the sub-feature data point group corresponding to the contour feature inspection result is as follows:

[0146] In the pending region, several fixed shape regions are randomly selected, and then the average frequency change speed and the average frequency change direction of the data points in these regions are calculated. In the calculation process, these regions move towards the direction close to the sub-feature data point group.

[0147] At this time, the trend line obtained should be a straight line or a line segment similar to a straight line. If the height of the line segment is within the allowed range, it means that the pending region is correlated with the sub-feature data point group corresponding to the contour feature inspection result. At this time, the pending region is divided into the sub-feature data point group corresponding to the contour feature inspection result. Otherwise, the original division mode is kept.

[0148] The application also provides an adaptive clustering device in a target detection process, comprising:

[0149] a data acquisition unit configured to acquire a data point group in a coverage range, the data point group being point cloud data generated in a unit time length;

[0150] a first classification unit configured to divide the data point group once to obtain at least one feature data point group;

[0151] The first checking unit is configured to check the speed consistency of the feature data point group, split the feature data point group when there are multiple speeds, and obtain multiple sub-feature data point groups after splitting the feature data point group;

[0152] The second checking unit is configured to assign a category label to the sub-feature data point group and perform contour feature checking according to the category label;

[0153] The second classification unit is configured to perform secondary division on the sub-feature data point group according to the contour feature checking result.

[0154] Further, the primary division on the data point group includes:

[0155] Calculating the frequency change direction of each data point in the data point group;

[0156] Performing primary screening on the data point group according to the frequency change direction, and removing invalid points in the data point group, the invalid points not having the frequency change direction;

[0157] Constructing an analysis grid using the data points, the number of edges of each grid in the analysis grid being the same;

[0158] Calculating the change trend of each grid on the time sequence, the change trend including an area change trend and a shape change trend;

[0159] Dividing the data point group according to the change trend to obtain at least one feature data point group;

[0160] The change trend of the grid associated with the feature data point group is the same.

[0161] Further, after the data point group is divided according to the change trend and at least one feature data point group is obtained, the method further includes:

[0162] Determining the shape of the feature data point group to obtain a contour feature;

[0163] Checking the data point group using the contour feature to determine whether there is a missing area in the data point group;

[0164] When there is a missing area in the data point group, marking the missing area as a to-be-assigned area.

[0165] Further, the method further includes:

[0166] Counting the occurrence frequency of the to-be-assigned area on the time sequence;

[0167] When the occurrence frequency of the to-be-assigned area is greater than or equal to a set frequency, determining the positional relationship between the to-be-assigned area and the feature data point group and calculating the distance between the to-be-assigned area and the millimeter wave detector according to the positional relationship;

[0168] selecting and scanning an auxiliary scanning area according to the position relationship and the distance;

[0169] When the area of the to-be-allocated region increases or remains unchanged, the to-be-allocated region is taken as a sub-feature data point group.

[0170] Further, the checking of the speed consistency of the feature data point group and the splitting processing of the feature data point group according to the checking result include:

[0171] selecting a plurality of data points in the feature data point group as representative points, the representative points being located at non-edge positions of the feature data point group;

[0172] calculating the frequency change speed and the frequency change direction of the representative points;

[0173] determining a splitting position according to the frequency change speed and the frequency change direction of the representative points;

[0174] performing secondary division of the data points at the splitting position using the splitting position.

[0175] Further, the secondary division of the data points at the splitting position using the splitting position includes:

[0176] selecting a data point region group at the splitting position as a marker data point region group;

[0177] determining a moving direction of the marker data point region group according to other data point region groups on both sides of the marker data point region group;

[0178] moving the marker data point region group in the moving direction and recording the average frequency change speed and the average frequency change direction of the marker data point region group;

[0179] when the average frequency change speed or the average frequency change direction of the marker data point region group changes, recording the change trend of the changed average frequency change speed or average frequency change direction;

[0180] determining a median position point of the change trend and sequentially connecting adjacent median position points to obtain a median position reference line;

[0181] performing secondary division of the data point group at the splitting position using the median position reference line.

[0182] Further, when the sub-feature data point group is divided according to the contour feature checking result, the method further includes:

[0183] determining a to-be-determined region according to the contour feature checking result;

[0184] Correlation degree of the to-be-determined region and the sub-feature data point group corresponding to the contour feature inspection result is calculated.

[0185] According to the correlation degree, the to-be-determined region is classified into the sub-feature data point group corresponding to the contour feature inspection result or remains in the original classification manner.

[0186] In one example, the units in any of the above apparatuses can be one or more integrated circuits configured to implement the above methods, for example, one or more application specific integrated circuits (ASICs), or, one or more digital signal processors (DSPs), or, one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0187] For another example, when the units in the apparatuses can be implemented in the form of a processing element scheduler, the processing element can be a general purpose processor, such as a central processing unit (CPU) or other processor that can invoke programs. For another example, these units can be integrated together to be implemented in the form of a system-on-a-chip (SOC).

[0188] In the present application, various objects such as messages / information / equipment / network elements / systems / apparatuses / actions / operations / processes / concepts, etc. that can appear in the present application are named. It can be understood that these specific names do not constitute a limitation on the related objects, and the assigned names can be changed according to the scene, context or usage habits, etc. The technical meaning of the technical terms in the present application should be mainly determined from the function and technical effect embodied / executed in the technical scheme.

[0189] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0190] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0191] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0192] Those skilled in the art can understand that the units and algorithm steps of the examples described in combination with the embodiments disclosed in the present application can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0193] It should also be understood that in various embodiments of the present application, the first, second, etc. are only used to represent different objects. For example, the first time window and the second time window are only used to represent different time windows. There should be no impact on the time window itself, and the above first, second, etc. should not limit the embodiments of the present application.

[0194] It should also be understood that in various embodiments of the present application, the terms and / or descriptions of different embodiments have consistency and can be mutually referred to if there is no special description and logical conflict. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0195] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a computer readable storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned computer readable storage medium includes: a U disk, a mobile hard disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0196] The present application also provides a millimeter wave detection machine, the millimeter wave detection machine comprising:

[0197] One or more memories for storing instructions;

[0198] One or more processors for calling and running the instructions from the memory, executing the methods as described above.

[0199] The present application also provides a computer program product comprising instructions which, when executed, cause the terminal device and the network device to perform the operations of the terminal device and the network device corresponding to the above-mentioned methods.

[0200] The present application also provides a chip system comprising a processor for implementing the functions involved in the above description, such as generating, receiving, sending, or processing the data and / or information involved in the above-mentioned methods.

[0201] The chip system can be composed of a chip, or it can include a chip and other discrete devices.

[0202] The processor mentioned in any of the above can be a CPU, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the program of the above-mentioned feedback information transmission method.

[0203] In a possible design, the chip system further includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and arranged on different devices, connected through wired or wireless means to support the chip system to realize various functions in the above embodiments. Alternatively, the processor and the memory can also be coupled on the same device.

[0204] Optionally, the computer instructions are stored in a memory.

[0205] Optionally, the memory is a storage unit within the chip, such as a register, a cache, etc. The memory can also be a storage unit outside the chip within the terminal, such as a ROM or other type of static storage device that can store static information and instructions, a RAM, etc.

[0206] It can be understood that the memory in the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.

[0207] The non-volatile memory can be a ROM, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM), or a flash memory.

[0208] The volatile memory can be a RAM used as an external cache. There are many different types of RAM, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synch link DRAM (SLDRAM), and direct Rambus RAM.

[0209] The embodiments of the present specific implementation are the preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, any equivalent changes made in the structure, shape, and principle of the present application should be covered within the protection scope of the present application.

Claims

1. An adaptive clustering method in a target detection process, characterized by, The method comprises the following steps: acquiring a data point group in a coverage range, the data point group being point cloud data generated in a unit time length; performing a first division on the data point group to obtain at least one feature data point group; checking the velocity consistency of the feature data point group, and performing a splitting process on the feature data point group when there are multiple velocities, the feature data point group being divided into multiple sub-feature data point groups after the splitting process; assigning a category label to the sub-feature data point group and performing a contour feature check according to the category label; performing a second division on the sub-feature data point group according to the contour feature check result; the first division on the data point group comprises the following steps: calculating the frequency change direction of each data point in the data point group; performing a first screening on the data point group according to the frequency change direction to remove invalid points in the data point group, the invalid points not having the frequency change direction; constructing an analysis grid using the data points, the number of edges of each grid in the analysis grid being the same; calculating the change trend of each grid in the time sequence, the change trend comprising an area change trend and a shape change trend; dividing the data point group according to the change trend to obtain at least one feature data point group; wherein the change trends of the grids associated with the feature data point group are the same; the second division on the sub-feature data point group according to the contour feature check result comprises the following steps: determining a pending area according to the contour feature check result; calculating the correlation degree of the pending area and the sub-feature data point group corresponding to the contour feature check result; selecting to divide the pending area into the sub-feature data point group corresponding to the contour feature check result or keep the original division mode according to the correlation degree.

2. The adaptive clustering method in target detection process according to claim 1, characterized in that, After the data point group is divided according to the change trend and at least one feature data point group is obtained, the method further comprises the following steps: determining the shape of the feature data point group to obtain a contour feature; checking the data point group using the contour feature to determine whether there is a missing area in the data point group; when there is a missing area in the data point group, marking the missing area as a to-be-assigned area.

3. The adaptive clustering method in target detection process according to claim 2, characterized in that, The method further comprises the following steps: statistically calculating the occurrence frequency of the to-be-assigned area in the time sequence; when the occurrence frequency of the to-be-assigned area is greater than or equal to a set frequency, determining the positional relationship between the to-be-assigned area and the feature data point group and calculating the distance between the to-be-assigned area and the millimeter wave detection machine according to the positional relationship; selecting an auxiliary scanning area according to the positional relationship and the distance and scanning the auxiliary scanning area; when the area of the to-be-assigned area increases or remains unchanged, regarding the to-be-assigned area as a sub-feature data point group.

4. The adaptive clustering method for object detection process of claim 1, wherein, The method of checking the velocity consistency of the feature data point group and performing a splitting process on the feature data point group according to the checking result comprises the following steps: selecting multiple data points in the feature data point group as representative points, the representative points being located at non-edge positions of the feature data point group; calculating the frequency change speed and the frequency change direction of the representative points; determining a splitting position according to the frequency change speed and the frequency change direction of the representative points; performing a second division on the data points at the splitting position using the splitting position.

5. The adaptive clustering method in target detection process according to claim 4, characterized in that, The second division on the data points at the splitting position using the splitting position comprises the following steps: selecting a data point area group at the splitting position as a marked data point area group; determining a moving direction of the marked data point region group according to other data point region groups on two sides of the marked data point region group; moving the marked data point region group in the moving direction and recording an average frequency change speed and an average frequency change direction of the marked data point region group; recording a change trend of the average frequency change speed or the average frequency change direction of the marked data point region group when the average frequency change speed or the average frequency change direction of the marked data point region group changes; determining median position points of the change trend and sequentially connecting adjacent median position points to obtain a median position reference line; performing secondary division on the data point group at the splitting processing position using the median position reference line.

6. An apparatus for adaptive clustering in a target detection process, characterized by The method comprises the following steps: a data acquisition unit is configured to acquire a data point group in a coverage range, the data point group being point cloud data generated in a unit time length; a first classification unit is configured to perform primary division on the data point group to obtain at least one feature data point group; a first checking unit is configured to check the speed consistency of the feature data point group, and perform splitting processing on the feature data point group when there are multiple speeds, so as to obtain multiple sub-feature data point groups after the splitting processing on the feature data point group; a second checking unit is configured to assign a category label to the sub-feature data point group and perform contour feature checking according to the category label; a second classification unit is configured to perform secondary division on the sub-feature data point group according to the contour feature checking result; the primary division on the data point group comprises the following steps: calculating a frequency change direction of each data point in the data point group; performing primary screening on the data point group according to the frequency change direction to remove invalid points in the data point group, the invalid points not having the frequency change direction; constructing an analysis grid using the data points, the number of edges of each grid in the analysis grid being the same; calculating a change trend of each grid in a time sequence, the change trend comprising an area change trend and a shape change trend; performing division on the data point group according to the change trend to obtain at least one feature data point group; wherein the change trends of the grids associated with the feature data point group are the same; the secondary division on the sub-feature data point group according to the contour feature checking result comprises the following steps: determining a pending region according to the contour feature checking result; calculating a correlation degree of the pending region and a sub-feature data point group corresponding to the contour feature checking result; selecting to divide the pending region into the sub-feature data point group corresponding to the contour feature checking result or keeping the original division mode according to the correlation degree.

7. A millimeter wave detection machine characterized by, The millimeter wave detection machine comprises: one or more memories configured to store instructions; one or more processors configured to call and run the instructions from the memories to perform the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises: a program configured to perform the method according to any one of claims 1 to 5 when the program is run by a processor.

Images