A target detection method and device, electronic equipment and medium

Abstract

Embodiments of the present application disclose a target detection method and device, electronic equipment and medium. The method comprises: if a range-doppler (RD) graph is not a first frame RD graph obtained by radar detection, predicting dense areas and sparse areas of target distribution in the RD graph according to a target detection result of a historical RD graph; for the dense areas, predicting non-target areas where no target exists, and determining a noise power estimation value of a to-be-detected cell in the dense areas according to at least one sub-area in the non-target areas, and performing target detection on the to-be-detected cell in the dense areas; for the sparse areas, predicting size information of a target corresponding to a to-be-detected cell, and determining a protection cell of the to-be-detected cell according to the size information; determining a reference cell according to the protection cell, determining a noise power estimation value of the to-be-detected cell in the sparse areas according to the reference cell, and performing target detection on the to-be-detected cell in the sparse areas. The technical solution can improve the efficiency and accuracy of target detection.

Description

Technical Field

[0001] This invention relates to the field of target detection technology, and in particular to a target detection method, apparatus, electronic device, and medium. Background Technology

[0002] With the booming development of the automotive industry, cars have become a common household item. While bringing convenience to people's travel, they have also caused problems such as environmental pollution and traffic congestion. In order to alleviate the impact of traditional vehicles on air pollution, the country has vigorously developed intelligent transportation in recent years. As a crucial component of intelligent transportation, sensing technology is essential for the smooth operation of vehicle systems. Millimeter-wave radar, with its low cost and other advantages, is increasingly being used in fields such as vehicle-mounted radar and traffic radar to achieve all-weather monitoring of people, vehicles, and other targets, compensating for the shortcomings of traditional sensing methods such as cameras.

[0003] Traditional millimeter-wave radar employs a Constant False Alarm Rate (CFAR) detector for target detection. CFAR is a common radar target detector, and its key characteristic is maintaining a constant false alarm probability, thus avoiding track overload caused by excessive false tracks. The basic idea of ​​this method is to assume that closer cells have more similar noise statistics. Data from cells near the cell to be detected is used as training samples to estimate the noise of the cell to be detected. The noise estimation power of the cell to be detected is calculated, and based on the estimated power of the cell to be detected in the radar image, the presence of the target is determined, thus achieving target detection.

[0004] However, the aforementioned methods face the problems of target fragmentation and multi-target detection in congested scenarios. Specifically, when the target itself is large (its size exceeds the radar range resolution), its energy typically occupies several consecutive range cells. To avoid target energy contaminating the noise energy estimation samples, traditional methods first select a few fixed cells near the target cell as guard cells, thus excluding cells containing target energy from the samples. Then, a certain amount of cell data near the guard cells is selected as training samples for noise energy statistics. This guard cell selection method leads to target fragmentation when detecting targets of different sizes, thereby increasing the computational load of target detection and reducing its accuracy.

[0005] Furthermore, in scenarios with low traffic volume and a wide range of vehicle speeds, the probability of vehicles being present near the target unit is low, allowing for accurate estimation of background noise and achieving good detection performance. However, during peak traffic periods such as mornings and evenings, holidays, or at traffic light intersections, traffic congestion is common. In these situations, multiple targets have similar positions, and their speeds are all distributed within a low range. In such scenarios, if the conventional CFAR detection algorithm is still used, the noise power estimation is easily overestimated due to the large number of targets near the target unit, resulting in missed detections of the target unit.

[0006] Therefore, how to effectively avoid the problem of large vehicle splitting caused by excessively large target size, and how to avoid background noise estimation errors caused by dense targets in congested scenarios, thereby improving the efficiency and accuracy of target detection, are urgent problems to be solved in the current field of target detection. Summary of the Invention

[0007] This invention provides a target detection method, apparatus, electronic device, and medium that can employ different noise power estimation strategies for dense and sparse target distribution areas, thereby improving the efficiency and accuracy of noise power estimation and contributing to improved target detection efficiency and accuracy.

[0008] According to one aspect of the present invention, a target detection method is provided, the method comprising:

[0009] If the range Doppler RD map is not the first frame RD map detected by the radar, then based on the target detection results of the historical RD maps, predict the dense and sparse areas of target distribution in the RD map.

[0010] For the dense region, predict the non-target region where no target exists, and determine the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region, and perform target detection on the unit to be detected in the dense region;

[0011] For the sparse region, the size information of the target corresponding to the unit to be detected is predicted, and the protection unit of the unit to be detected is determined based on the size information;

[0012] A reference unit is determined based on the protection unit, and a noise power estimate of the unit to be detected in the sparse region is determined based on the reference unit. Target detection is then performed on the unit to be detected in the sparse region.

[0013] According to another aspect of the present invention, a target detection device is provided, comprising:

[0014] The region type prediction module is used to predict the dense and sparse regions of target distribution in the RD map based on the target detection results of the historical RD map if the range Doppler RD map is not the first frame RD map detected by the radar.

[0015] The dense area target detection module is used to predict non-target areas where no target exists in the dense area, and determine the noise power estimate of the unit to be detected in the dense area based on at least one sub-region in the non-target area, and perform target detection on the unit to be detected in the dense area.

[0016] The protection unit determination module is used to predict the size information of the target corresponding to the unit to be detected for the sparse region, and determine the protection unit of the unit to be detected based on the size information.

[0017] The sparse region target detection module is used to determine a reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region.

[0018] According to another aspect of the present invention, a target detection electronic device is provided, the electronic device comprising:

[0019] At least one processor; and

[0020] A memory communicatively connected to the at least one processor; wherein,

[0021] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the target detection method according to any embodiment of the present invention.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the target detection method according to any embodiment of the present invention.

[0023] The technical solution of this invention, if the range Doppler RD map is not the first frame RD map obtained by radar detection, predicts the dense and sparse regions of target distribution in the RD map based on the target detection results of historical RD maps; for dense regions, predicts non-target regions where no targets exist, and determines the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region, and performs target detection on the unit to be detected in the dense region; for sparse regions, predicts the size information of the target corresponding to the unit to be detected, and determines the protection unit of the unit to be detected based on the size information; determines the reference unit based on the protection unit, determines the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and performs target detection on the unit to be detected in the sparse region. This technical solution can adopt different noise power estimation strategies for dense and sparse target distribution regions, avoiding the problem of large noise power estimation errors caused by large vehicle splitting and target congestion problems, improving the efficiency and accuracy of noise power estimation, and helping to improve the efficiency and accuracy of target detection.

[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of a target detection method provided according to Embodiment 1 of the present invention;

[0027] Figure 2 This is a schematic diagram of a target distribution area type provided in Embodiment 1 of the present invention;

[0028] Figure 3a This is a first schematic diagram of a sub-region in a non-target region provided in Embodiment 1 of the present invention;

[0029] Figure 3b This is a second schematic diagram of a sub-region in a non-target region provided in Embodiment 1 of the present invention;

[0030] Figure 4 This is a schematic diagram of noise power estimation in a sparse region according to Embodiment 1 of the present invention;

[0031] Figure 5 This is a flowchart of a target vehicle detection method provided in Embodiment 1 of the present invention;

[0032] Figure 6 This is a flowchart of a target detection method provided according to Embodiment 2 of the present invention;

[0033] Figure 7 This is a flowchart of a target detection method provided in Embodiment 3 of the present invention;

[0034] Figure 8 This is a schematic diagram of the structure of a target detection device according to Embodiment 4 of the present invention;

[0035] Figure 9 This is a schematic diagram of the structure of an electronic device that implements a target detection method according to an embodiment of the present invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] Example 1

[0039] Figure 1 This is a flowchart of a target detection method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations requiring rapid and accurate target detection. The method can be executed by a target detection device, which can be implemented in hardware and / or software and can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0040] S110, if the range Doppler RD map is not the first frame RD map detected by the radar, then based on the target detection results of the historical RD maps, predict the dense and sparse areas of target distribution in the RD map.

[0041] The range-Doppler (RD) map refers to the distribution of radar echo energy in the range and Doppler two-dimensional planes after range and Doppler processing, serving as the basis for radar target detection. Specifically, after the radar receives the radar echo signal reflected from a target, it performs range Fourier transform and Doppler Fourier transform on the radar echo signal to obtain the range-Doppler RD map. The first frame RD map refers to the first frame of range-Doppler RD map detected after the radar is powered on. Historical RD maps refer to range-Doppler RD maps detected by the radar before the current range-Doppler RD map. Target detection results can be used to indicate whether a target exists in the RD map and the target's location within the RD map. Dense regions can be used to characterize areas with a high concentration of targets. Sparse regions can be used to characterize areas with a low concentration of targets. A target can be understood as the object to be detected, such as a vehicle or pedestrian.

[0042] In this embodiment, if the current range-Doppler RD map is the first frame RD map detected by the radar, target detection can be performed according to existing conventional methods. If the current range-Doppler RD map is not the first frame RD map detected by the radar, the target can be clustered and tracked based on the target detection results of historical RD maps, such as the target's position in the historical RD maps and the target's speed and direction of movement. Then, based on the target distribution, the dense and sparse areas of target distribution in the current RD map can be predicted.

[0043] Figure 2 This is a schematic diagram of a target distribution area type provided in Embodiment 1 of the present invention. The solid rectangular frame represents the current RD map, and the solid circle represents the target. Figure 2 As shown, the horizontal axis of the RD plot represents Doppler, the vertical axis represents distance, and the 0 Doppler region represents the area where the Doppler zero point is located. The current RD plot includes multiple targets. In the current RD plot, the area below the dashed line is the dense area of ​​the predicted target distribution, where there are more targets; the area above the dashed line is the sparse area of ​​the predicted target distribution, where there are fewer targets.

[0044] S120: For dense areas, predict non-target areas where no target exists, and determine the noise power estimate of the unit to be detected in the dense area based on at least one sub-region in the non-target area, and perform target detection on the unit to be detected in the dense area.

[0045] Here, "non-target region" can refer to a region within a dense region where no target exists. "Sub-region" can refer to a portion of a non-target region. "Detection unit" can refer to a unit in the RD diagram awaiting target detection, specifically the detection unit corresponding to a unit distance and unit Doppler in the RD diagram. "Noise power estimate" can refer to the predicted noise power obtained by estimating the noise power of the detection unit.

[0046] In this embodiment, after predicting the dense region of target distribution in the RD map, non-target regions where no target exists in the dense region can be predicted directly based on the target distribution. Then, at least one sub-region is determined in the non-target region, and the noise power estimate of the unit to be detected in the dense region is determined based on at least one sub-region in the non-target region, so as to perform target detection on the unit to be detected in the dense region. Figure 3a This is a first schematic diagram of a sub-region within a non-target region provided in Embodiment 1 of the present invention. For example, as shown... Figure 3a As shown, a sub-region, i.e., region 1, can be selected from the non-target region where no target exists in the dense region, and the average noise power of region 1 can be used as the noise power estimate of the unit to be detected. Here, m1 and m2 represent the distance coordinates of the lower and upper boundaries of region 1, respectively, and n... l1 and n l2 These represent the Doppler coordinates of the left and right boundaries of region 1, respectively. Figure 3b This is a second schematic diagram of a sub-region within a non-target region provided in Embodiment 1 of the present invention. For example... Figure 3b As shown, two sub-regions are selected within the dense region, designated as region 1 and region 2. Here, m1 and m2 represent the distance coordinates between the lower and upper boundaries of the two sub-regions, respectively, and n... l1 and n l2 Let n represent the Doppler coordinates of the left and right boundaries of region 1, respectively. r1 and n r2The left and right boundaries of zone 2 are represented by their Doppler coordinates, respectively. Furthermore, zones 1 and 2 are two rectangular regions of equal size, located within the same distance range. When determining the noise power estimate of the target unit in a dense region, the average noise power of zones 1 and 2 can be calculated separately, and the smaller value can be selected as the noise power estimate of the target unit. The above is just an example; there is no limit to the number of sub-regions selected. When there are at least two sub-regions, the minimum average noise power of each sub-region can be used as the noise power estimate of the target unit. It should be noted that in congested traffic conditions, target groups typically cluster within a certain distance range and in low-Doppler regions, while targets in high-Doppler ranges are rare. Traffic radar background noise is primarily receiver thermal noise, exhibiting good uniformity within the RD map, especially for different Doppler units within the same distance cell. Since the receiver noise power remains almost constant within a frame, it exhibits high uniformity within the Doppler interval. To avoid the influence of targets on noise power estimation, the selection of the Doppler range should avoid the Doppler intervals where congested targets are located. Therefore, a distance range corresponding to the target group's distribution distance and the positive and negative high Doppler range can be selected from the non-target areas of the dense region as the basis for determining the sub-region. In this embodiment, no limitations are placed on the size, location, and shape of the sub-region, which can be set according to actual needs.

[0047] S130: For sparse regions, predict the size information of the target corresponding to the unit to be detected, and determine the protection unit of the unit to be detected based on the size information.

[0048] Size information can be used to characterize the size of the target, describing the number of units the target occupies in the RD map. Specifically, size information can include size information in the range direction and size information in the Doppler direction. A protective unit can refer to the outer unit of the unit to be detected. In this embodiment, for sparse regions, the target image corresponding to the first frame of the RD map can be acquired and recognized by a radar-linked camera to determine the target's size information. Then, the size information of the target corresponding to the unit to be detected in the current RD map can be predicted through target tracking. Prediction can also be made based on the target type at the corresponding position of the unit to be detected in the previous frame of the RD map. For example, if the target type at unit 1 in the previous frame of the RD map is a large vehicle, and trajectory prediction determines that it will move to unit 2, then for the current frame of the RD map, the target corresponding to unit 2 is the same as the target at unit 1 in the previous frame of the RD map, and the target type is a large vehicle.

[0049] After predicting the size information of the target corresponding to the unit to be detected, the protection units of the unit to be detected can be determined based on the size information, thereby ensuring that the target corresponding to the unit to be detected can be completely contained within the range of the protection units. For example, the number of protection units can be set to be positively correlated with the size information of the target corresponding to the unit to be detected. If the size information of the target corresponding to the unit to be detected is greater than a preset threshold, it is determined that the number of protection units of the unit to be detected exceeds the preset threshold; if the size information of the target corresponding to the unit to be detected is less than or equal to the preset threshold, it is determined that the number of protection units of the unit to be detected is less than the preset threshold. In this embodiment, the protection units can be set in the neighborhood of the unit to be detected, surrounding the unit to be detected, such as... Figure 4 As shown, by determining the protection unit based on the size information of the target corresponding to the unit to be detected, the range of the protection unit can be closer to the actual size of the target, thereby protecting the entire target and preventing the target from being fragmented, thus solving the problem of a large target being misidentified as multiple small targets.

[0050] S140, determine the reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region.

[0051] Here, the reference unit can refer to the peripheral unit of the protection unit. For example, a reference unit is selected within the neighboring units outside the protection unit. The number of reference units can be determined according to the actual situation, for example, it can be 50. In this embodiment, the corresponding reference units are first determined as samples for noise power estimation based on the protection unit. Figure 4 This is a schematic diagram illustrating noise power estimation in a sparse region according to Embodiment 1 of the present invention. Figure 4 As shown, assume that the sample sets on both sides of the distance dimension of the unit to be detected are Γ. u ={x i ,i∈Ι u} and Γ l ={x i ,i∈Ι l}, where Γ u and Γ l Let I represent the upper training sample (high-range gate training sample) and lower training sample (low-range gate training sample) of the unit to be detected, respectively. u and I l These represent the upper training sample index set and the lower training sample index set of the unit to be detected, respectively.

[0052] Unit samples whose distance values ​​are equal to those of the unit to be detected in the same distance dimension, i.e. Figure 4 The unit samples in the shaded area shown can be selected in half of the Γ area. u In the middle, the other half is in Γl In the middle. For example, such as Figure 4 As shown, the unit samples located in the shaded region to the left of the unit to be detected are assigned to the low-range gate training samples Γ. l The unit samples located in the shaded region to the right of the unit to be detected are assigned to the high-distance gate training samples Γ. u The noise power estimates for both sides are calculated separately. and in, and Let represent the noise power estimates of the high-range gate training samples and the low-range gate training samples, respectively, where N is the number of training samples on one side. Then, the minimum value is selected as the noise power estimate σ for the unit to be detected. 2 ,Right now Then, based on the noise power estimate of the unit to be detected, target detection is performed on the unit to be detected in the sparse region.

[0053] The technical solution of this invention, if the range Doppler RD map is not the first frame RD map obtained by radar detection, predicts the dense and sparse regions of target distribution in the RD map based on the target detection results of historical RD maps; for dense regions, predicts non-target regions where no targets exist, and determines the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region, and performs target detection on the unit to be detected in the dense region; for sparse regions, predicts the size information of the target corresponding to the unit to be detected, and determines the protection unit of the unit to be detected based on the size information; determines the reference unit based on the protection unit, determines the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and performs target detection on the unit to be detected in the sparse region. This technical solution can adopt different noise power estimation strategies for dense and sparse target distribution regions, avoiding the problem of large noise power estimation errors caused by large vehicle splitting and target congestion problems, improving the efficiency and accuracy of noise power estimation, and helping to improve the efficiency and accuracy of target detection.

[0054] In this embodiment, optionally, the method further includes: performing clustering processing on the target detection results of the unit to be detected to obtain the target in the RD map; tracking the target in the RD map to determine the predicted position of the target in the RD map; determining the number of tracks of each distance unit in the RD map based on the predicted position, and determining the region type to which the distance unit belongs based on the number of tracks, so as to predict the target distribution in the RD map of the next frame; wherein, the region type includes dense region and sparse region.

[0055] The region type can be used to characterize the density of target distribution. Specifically, the region type can include dense regions and sparse regions. In this embodiment, after target detection is performed on the units to be detected in the RD map, the target detection results of the units to be detected can be further clustered to determine the targets. Specifically, after target detection is performed on the units to be detected, if a target exists at the unit, the third dimension value corresponding to the unit (excluding distance and Doppler dimension) is assigned a value of 1; if no target exists at the unit, the third dimension value corresponding to the unit (excluding distance and Doppler dimension) is assigned a value of 0, forming a three-dimensional result map. The three-dimensional result map is then clustered to determine the targets. Then, the targets are tracked to determine their predicted positions. The RD map is divided into different distance gates according to the distance dimension, where a distance gate represents a Doppler unit corresponding to the same distance value in the distance dimension. For each distance gate in the RD map, the total number of targets whose predicted positions fall within each distance gate is taken as the number of tracks Q(m) within the corresponding distance gate. A threshold of q is pre-set for dense target regions. When Q(m) > q, the m-th distance gate is determined to be a dense target region; otherwise, it is determined to be a sparse target region. Thus, the dense region can be determined based on the set of dense target regions to predict the target distribution in the RD map of the next frame.

[0056] This scheme, through this setting, determines the region type of each distance cell in the RD map based on the target detection results of the unit to be detected, and through clustering and target tracking, can better predict the target distribution in the RD map of the next frame, thus improving the accuracy of target distribution prediction.

[0057] In this embodiment, optionally, the method further includes: determining the type of the target based on the clustering processing result, so as to predict the type of the target corresponding to the unit to be detected in the next frame RD image based on the type of the target and the predicted position of the target.

[0058] The target type can be used to reflect the size of the target. Taking a vehicle as an example, the target type is the vehicle type. For instance, vehicle types can be roughly divided into large vehicles or small vehicles, or they can be more precisely divided into buses, trucks, off-road vehicles, SUVs, or sedans, etc. This embodiment does not impose any limitations on the classification of target types and can be set according to actual application needs.

[0059] In this embodiment, after target detection is performed on the unit to be detected, if a target exists at the unit, the third dimension value (excluding distance and Doppler dimension) corresponding to the unit is assigned a value of 1; if no target exists at the unit, the third dimension value (excluding distance and Doppler dimension) is assigned a value of 0, forming a three-dimensional result image. Clustering is then performed on the three-dimensional result image to determine the target. The target type is determined based on the number of units occupied by the target. For example, if the number of units occupied by the target exceeds a first threshold, the target type is determined to be a large target; if the number of units occupied by the target is less than a second threshold, the target type is determined to be a small target; and the remaining target types are medium targets. Furthermore, before target detection is performed on the next frame RD image, the type of the target corresponding to the unit to be detected in the next frame RD image can be predicted based on the target type and the predicted target position. For example, assuming the target type corresponding to the unit to be detected in the current frame RD image is a large target located at position one, and the predicted position is position two in the next frame RD image, then the target type corresponding to the unit to be detected at position two in the next frame RD image is a large target.

[0060] In this embodiment, if at least two targets are identified as vehicles, their relative positions in at least two RD (Real-Time Analysis) frames can be determined based on their predicted positions. If the distance between at least two targets is less than a preset distance in both RD frames, and their relative positions remain unchanged, then the at least two targets are determined to be a single target obtained by segmenting a large vehicle. Therefore, the at least two targets are merged into one target, identified as a large vehicle, and further processed. The preset distance can be set according to actual conditions; for example, it can be set as the distance between two vehicles when they are relatively stationary under normal circumstances, such as the distance between two vehicles traveling at the same speed.

[0061] In this embodiment of the application, for a detectable unit whose target type cannot be predicted from the RD image of the previous frame, it can be processed according to the default type, for example, the target type corresponding to the detectable unit is a small target by default.

[0062] This scheme, through this setting, can predict the type of the target corresponding to the unit to be detected in the next frame RD image based on the type of the target and the predicted position of the target, thereby improving the accuracy of target type prediction.

[0063] In this embodiment, the target type can reflect the target's size information. For example, if the target type is a large truck, the target size can be determined as that of a large truck; if the target type is a small car, the target size can be determined as that of a small car. This facilitates the subsequent adaptive determination of the protection unit size of the unit to be detected based on the target's size information, thus protecting the target and solving the problem of splitting a large target into smaller targets. If the target type corresponding to the unit to be detected is not predicted before target detection, the target's size information cannot be determined in a timely and accurate manner, and the aforementioned adaptive determination of the protection unit size cannot be achieved. However, in this embodiment, based on the target type and the predicted position of the target, the target type corresponding to the unit to be detected in the next frame RD image is predicted. When detecting the unit to be detected in the next frame RD, the target type corresponding to the unit to be detected can be determined in a timely and accurate manner, facilitating the subsequent determination of the target's size information based on the type, adaptively determining the protection unit size, and solving the problem of splitting a large target into smaller targets.

[0064] Figure 5 This is a flowchart of a target vehicle detection method provided in Embodiment 1 of the present invention. Figure 5 As shown, the process first selects the unit to be detected and determines whether the application scenario in which the unit is located is a dense or sparse scenario. In a dense scenario, the corresponding dense target detection strategy is adopted. This involves predicting non-target regions where no target exists within the dense region, and determining the noise power estimate of the unit to be detected in the dense region based on at least one sub-region within the non-target region. Target detection is then performed on the unit to be detected in the dense region. In a sparse scenario, the corresponding sparse target detection strategy is adopted. First, it is determined whether the target vehicle is a large vehicle or a small vehicle. If it is a large vehicle, a large vehicle detection strategy is adopted. This involves predicting the size information of the target corresponding to the unit to be detected in the sparse region, determining the protection unit of the unit to be detected based on the size information, determining the reference unit based on the protection unit, and determining the noise power estimate of the unit to be detected in the sparse region based on the reference unit. Target detection is then performed on the unit to be detected in the sparse region. If it is a small vehicle, a traditional fixed protection unit strategy is used for target detection. Based on the target detection results, clustering and target tracking are performed sequentially to obtain the target tracking results. Then, based on the target tracking results, scene identification and target classification are performed to predict the target distribution and the type of the target corresponding to the unit to be detected in the RD image of the next frame.

[0065] Example 2

[0066] Figure 6This is a flowchart of a target detection method provided in Embodiment 2 of the present invention. This embodiment is based on the above embodiment and optimized. Specifically, the optimization is as follows: determining the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region, including: predicting the distance range of the target distribution in the dense region and determining the candidate region in the non-target region corresponding to the distance range; selecting at least one sub-region from the candidate regions; determining the energy mean in each sub-region, and taking the minimum value among the energy mean values ​​as the noise power estimate of the unit to be detected in the dense region.

[0067] like Figure 6 As shown, the method in this embodiment specifically includes the following steps:

[0068] S210, if the range Doppler RD map is not the first frame RD map detected by the radar, then based on the target detection results of the historical RD maps, predict the dense and sparse areas of target distribution in the RD map.

[0069] S220, for dense areas, predict non-target areas where no target exists, predict the distance range of target distribution in dense areas, and determine candidate areas in non-target areas corresponding to the distance range.

[0070] In this context, a candidate region can refer to a region outside the target region that corresponds to the distance range of the target distribution. The distance range of the target distribution can refer to the distance range encompassing all targets within the dense region. In this embodiment, for a dense region, firstly, non-target regions where no targets exist are predicted, and the distance range of the target distribution within the dense region is predicted. Then, regions outside the target region that correspond to the distance range of the target distribution are determined as candidate regions. In other words, the distance range of the candidate regions is consistent with the distance range of the target distribution.

[0071] S230, Select at least one sub-region from the candidate regions.

[0072] In this embodiment, at least one sub-region is arbitrarily selected from the candidate regions. The shape and size of this at least one sub-region can be the same or different. To make the noise power estimate more accurate, at least one sub-region of the same size and shape can be selected within the high Doppler range on both sides of the 0 Doppler region.

[0073] S240, determine the average energy value in each sub-region, and use the minimum value among the average energy values ​​as the noise power estimate of the unit to be detected in the dense region, and perform target detection on the unit to be detected in the dense region.

[0074] Here, the average energy can refer to the average energy of a sub-region. Taking Figure 3 as an example, firstly, using the formula... Determine the average energy value of zone 1 using the formula. Determine the average energy value of region 2. Among them, and Let represent the energy mean of region 1 and region 2, respectively. Let x(m,n) represent the echo energy of each cell in the RD diagram, where m and n represent the distance cell index and the Doppler cell index, respectively. Then, the minimum energy mean is selected from the energy mean values ​​of regions 1 and 2 as the noise power estimate of the cell to be detected in the dense region. Right now Target detection of units to be detected in dense areas is performed based on noise power estimates.

[0075] S250, for sparse regions, predicts the size information of the target corresponding to the unit to be detected, and determines the protection unit of the unit to be detected based on the size information.

[0076] S260, determine the reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region.

[0077] The technical solution of this invention predicts the distance range of target distribution in a dense region and determines candidate regions in non-target regions corresponding to this distance range; selects at least one sub-region from the candidate regions; determines the average energy value in each sub-region, and uses the minimum value among the average energy values ​​as the noise power estimate of the unit to be detected in the dense region. This technical solution can employ different noise power estimation strategies for dense and sparse target distribution regions, and can determine sub-regions based on the predicted distance range of target distribution in non-target regions and dense regions to determine the noise power estimate of the unit to be detected in the dense region. This avoids the problem of large noise power estimation errors caused by large vehicle splitting and target congestion problems, further improving the efficiency and accuracy of noise power estimation, and contributing to improved efficiency and accuracy of target detection.

[0078] In this embodiment, optionally, target detection of the unit to be detected in a dense area includes: determining the signal-to-noise ratio based on the ratio of the energy value of the unit to be detected to the noise power estimate; determining a threshold value based on a preset false alarm rate and the number of reference units in the sub-region corresponding to the noise power estimate; and determining the target detection result of the unit to be detected based on the comparison result of the signal-to-noise ratio and the threshold value.

[0079] The preset false alarm rate can refer to a pre-set false alarm rate. The threshold value can refer to a pre-set threshold used to measure the signal-to-noise ratio (SNR). Taking Figure 3 as an example, based on the above example, assuming the current unit to be detected is (m0, n0), and its corresponding echo energy value is x(m0, n0), the SNR is first determined based on the ratio of the energy value of the unit to be detected to the noise power estimate. Assume the preset false alarm rate is p fa Based on the sizes of regions 1 and 2, the number of reference units in both sub-regions can be determined to be N1 = (m2 - m1 + 1)(n l2 -n l1 +1), from which the exit limit value can be determined. The signal-to-noise ratio (SNR) is then compared with a threshold value, and the target detection result for the unit to be detected is determined based on the comparison result. Specifically, the target is determined to exist when SNR ≥ α, and the target is determined to not exist when SNR < α.

[0080] This solution, through this configuration, can quickly and accurately determine the target detection result of the unit to be detected based on the comparison result of the signal-to-noise ratio and the threshold value.

[0081] Example 3

[0082] Figure 7 This is a flowchart of a target detection method provided in Embodiment 3 of the present invention. This embodiment is based on the above embodiment and optimized. Specifically, the optimization is as follows: predicting the size information of the target corresponding to the unit to be detected, and determining the protection unit of the unit to be detected based on the size information, including: predicting the type of the target corresponding to the unit to be detected, and determining the size information corresponding to the type; determining the number of protection units in the range dimension based on the size information and the radar range resolution; determining the number of protection units in the Doppler dimension based on the preset target motion acceleration, the time interval between two adjacent RD images, and the radar velocity resolution; and determining protection units in the neighborhood units of the unit to be detected based on the number of protection units in the range dimension and the number of protection units in the Doppler dimension.

[0083] like Figure 7 As shown, the method in this embodiment specifically includes the following steps:

[0084] S310: If the range Doppler RD map is not the first frame RD map detected by the radar, then based on the target detection results of the historical RD maps, predict the dense and sparse areas of target distribution in the RD map.

[0085] S320: For dense areas, predict non-target areas where no target exists, and determine the noise power estimate of the unit to be detected in the dense area based on at least one sub-region in the non-target area, and perform target detection on the unit to be detected in the dense area.

[0086] S330, for sparse regions, predicts the type of the target corresponding to the unit to be detected and determines the size information corresponding to the type.

[0087] Taking vehicles as an example, for sparse areas, the vehicle type corresponding to the unit to be detected can be predicted by target tracking based on the target detection results of historical RD maps. That is, the predicted position of whether the unit to be detected is a large vehicle or a small vehicle can be determined, and the size information corresponding to the vehicle type can be determined at the same time.

[0088] S340 determines the number of protection units in the range dimension based on size information and radar range resolution.

[0089] Here, range resolution can refer to the smallest distance that the radar can identify. In this embodiment, optionally, the number of protection units in the range dimension is determined based on the size information and the radar's range resolution, including: predicting the number of range units occupied by the target based on the ratio of the size information to the range resolution; wherein, a range unit is the smallest unit in the range dimension; and determining the number of protection units in the range dimension based on the number of range units with main lobe broadening after Fourier windowing processing in the range dimension, and the predicted number of range units occupied by the target.

[0090] For example, suppose the typical dimensions of the large vehicle and the small vehicle are r, respectively. b and r s Based on the effective bandwidth B of the radar signal, the radar's range resolution can be calculated as follows: Where c is the speed of light. Furthermore, based on the ratio of size information to distance resolution, the number of distance cells occupied by the target can be predicted. If the target is a large vehicle, the number of distance cells occupied by the large vehicle in the RD map can be N. rb =[r b / Δr] even If the target is a car, then the number of distance cells occupied by the car in the RD map can be calculated as N. rs =[r s / Δr] even ,in[·] even This indicates rounding down to the nearest even number.

[0091] Considering that Fourier windowing of the distance dimension will cause the main lobe to widen by one distance unit, the number of protection units in the distance dimension on one side of the vehicle can be determined as M. rb =N rb / 2+1, and determine the number of protection units in the single-sided distance dimension of the trolley as M. rs =N rs / 2+1. Therefore, the number of protection units in the distance dimension of the large vehicle is 2M. rb =N rb +2, the number of protection units in the distance dimension of the vehicle is 2M. rs =N rs +2.

[0092] This scheme determines the number of protection units in the distance dimension by taking into account the number of main lobe widening distance units after Fourier windowing processing in the distance dimension and the number of distance units occupied by the predicted target. This improves the accuracy of the number of protection units in the distance dimension by considering the impact of Fourier windowing processing in the distance dimension.

[0093] S350 determines the number of protection units in the Doppler dimension based on the preset acceleration of the target motion, the time interval between two adjacent RD images, and the radar velocity resolution.

[0094] Wherein, velocity resolution can refer to the minimum velocity that the radar can identify. In this embodiment, optionally, the number of protection units in the Doppler dimension is determined based on the preset target motion acceleration, the time interval between two adjacent RD images, and the radar velocity resolution. This includes: predicting the target velocity change based on the preset target motion acceleration and the time interval between two adjacent RD images; predicting the number of Doppler units occupied by the target based on the ratio of the velocity change to the velocity resolution; and determining the number of protection units in the Doppler dimension based on the number of main lobe-expanded Doppler units after Fourier windowing processing in the Doppler dimension and the predicted number of Doppler units occupied by the target.

[0095] For example, assuming the preset acceleration of the target motion is 'a', the time interval between two adjacent RD frames is 'T', and the radar velocity resolution is 'a'. Where λ is the radar wavelength. Based on this, the change in the target's velocity can be calculated as Δv1 = aT, and further, the number of Doppler cells occupied by the target in the RD diagram can be calculated as N. v =[Δv1 / Δv] even Considering that Fourier windowing in the Doppler dimension will cause the main lobe to widen by one Doppler unit, the number of protection units in the single-sided Doppler dimension of the target can be determined as M. v =N v / 2+1, meaning the number of protection units in the Doppler dimension is 2M. v =N v +2.

[0096] In summary, based on the target type, an appropriate protection unit selection strategy is adopted. Taking a vehicle as the target, if the vehicle type is a large vehicle, then the number of protection units selected in the single-sided distance dimension is M. rb If the vehicle type is a car or there is no vehicle in the unit to be detected, then select M as the number of protection units in the single-sided distance dimension. rs The number of protection units in the single-sided Doppler dimension is uniformly M. v .

[0097] This scheme determines the number of protection units in the Doppler dimension based on the number of main lobe-widened Doppler units after Fourier windowing processing in the Doppler dimension and the number of Doppler units occupied by the predicted target. This takes into account the impact of Fourier windowing processing in the Doppler dimension and improves the accuracy of the number of protection units in the Doppler dimension.

[0098] S360 determines the protection element in the neighborhood of the element to be detected based on the number of protection elements in the distance dimension and the number of protection elements in the Doppler dimension.

[0099] In this context, a neighboring unit can refer to an outer unit adjacent to the unit to be detected. In this embodiment, after determining the number of protection units in the distance dimension and the number of protection units in the Doppler dimension, protection units can be further determined in the neighboring units of the unit to be detected.

[0100] S370: Determine the reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region.

[0101] The technical solution of this invention predicts the type of target corresponding to the unit to be detected and determines the size information corresponding to the type; based on the size information and the radar's range resolution, it determines the number of protection units in the range dimension; based on the preset target motion acceleration, the time interval between two adjacent RD images, and the radar's velocity resolution, it determines the number of protection units in the Doppler dimension; based on the number of protection units in the range dimension and the number of protection units in the Doppler dimension, it determines protection units in the neighborhood units of the unit to be detected. This technical solution can adopt different noise power estimation strategies for dense and sparse target distribution areas, and quickly and accurately determine the number of protection units based on the radar resolution, avoiding the problem of large noise power estimation errors caused by large vehicle splitting and target congestion problems, further improving the efficiency and accuracy of noise power estimation, and helping to improve the efficiency and accuracy of target detection.

[0102] Example 4

[0103] Figure 8 This is a schematic diagram of a target detection device provided in Embodiment 4 of the present invention. This device can execute the target detection method provided in any embodiment of the present invention, and possesses the corresponding functional modules and beneficial effects for executing the method. For example... Figure 8 As shown, the device includes:

[0104] The region type prediction module 410 is used to predict the dense and sparse regions of target distribution in the RD map based on the target detection results of the historical RD map if the range Doppler RD map is not the first frame RD map detected by the radar.

[0105] The dense area target detection module 420 is used to predict non-target areas where no target exists in the dense area, and determine the noise power estimate of the unit to be detected in the dense area based on at least one sub-region in the non-target area, and perform target detection on the unit to be detected in the dense area.

[0106] The protection unit determination module 430 is used to predict the size information of the target corresponding to the unit to be detected for the sparse region, and determine the protection unit of the unit to be detected based on the size information.

[0107] The sparse region target detection module 440 is used to determine a reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region.

[0108] Optionally, the dense area target detection module 420 is used for:

[0109] Predict the distance range of the target distribution in the dense region, and determine the candidate region corresponding to the distance range in the non-target region;

[0110] Select at least one sub-region from the candidate regions;

[0111] The average energy value of each sub-region is determined, and the minimum value among the average energy values ​​is used as the noise power estimate of the unit to be detected in the dense region.

[0112] Optionally, the dense area target detection module 420 is further configured to:

[0113] The signal-to-noise ratio is determined based on the ratio of the energy value of the unit to be detected to the estimated noise power value.

[0114] The threshold value is determined based on the preset false alarm rate and the number of reference cells in the sub-region corresponding to the noise power estimate;

[0115] Based on the comparison result between the signal-to-noise ratio and the threshold value, the target detection result for the unit to be detected is determined.

[0116] Optionally, the protection unit determination module 430 includes:

[0117] A size information determination unit is used to predict the type of the target corresponding to the unit to be detected, and determine the size information corresponding to the type;

[0118] The first protection unit quantity determination unit is used to determine the number of protection units in the range dimension based on the size information and the range resolution of the radar.

[0119] The second protection unit is used to determine the number of protection units in the Doppler dimension based on the preset target motion acceleration, the time interval between two adjacent RD images, and the radar velocity resolution.

[0120] A protection unit is used to determine protection units in the neighborhood units of the unit to be detected based on the number of protection units in the distance dimension and the number of protection units in the Doppler dimension.

[0121] Optionally, the first protection unit quantity determination unit is specifically used for:

[0122] Based on the ratio of the size information to the distance resolution, the number of distance units occupied by the target is predicted; wherein, the distance unit is the smallest unit in the distance dimension;

[0123] The number of protection units in the distance dimension is determined based on the number of main lobe widening distance units after Fourier windowing processing in the distance dimension, and the number of distance units occupied by the predicted target.

[0124] Optionally, the second protection unit quantity determination unit is specifically used for:

[0125] Based on the preset acceleration of the target motion and the time interval between two adjacent RD images, predict the change in the target's velocity.

[0126] The number of Doppler cells occupied by the target is predicted based on the ratio of the velocity change to the velocity resolution.

[0127] The number of protection cells in the Doppler dimension is determined based on the number of main lobe-widened Doppler cells after Fourier windowing in the Doppler dimension, and the number of Doppler cells occupied by the predicted target.

[0128] Optionally, the device further includes:

[0129] The target detection module is used to perform clustering processing on the target detection results of the unit to be detected to obtain the targets in the RD graph;

[0130] The target prediction location determination module is used to track the target in the RD map and determine the predicted location of the target in the RD map;

[0131] The region type determination module is used to determine the number of tracks of each distance cell in the RD map based on the predicted location, and to determine the region type to which the distance cell belongs based on the number of tracks, so as to predict the target distribution in the next frame of the RD map; wherein, the region type includes dense region and sparse region.

[0132] Optionally, the device further includes:

[0133] The target type determination module is used to determine the type of the target based on the clustering processing results, so as to predict the type of the target corresponding to the unit to be detected in the next frame RD map based on the type of the target and the predicted position of the target.

[0134] The target detection device provided in this embodiment of the invention can execute a target detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0135] Example 5

[0136] Figure 9 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0137] like Figure 9 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0138] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0139] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as object detection methods.

[0140] In some embodiments, the target detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the target detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the target detection method by any other suitable means (e.g., by means of firmware).

[0141] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0143] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0144] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0145] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0146] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0147] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0148] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A target detection method, characterized in that, The method includes: If the range Doppler RD map is not the first frame RD map detected by the radar, then based on the target detection results of the historical RD maps, predict the dense and sparse areas of target distribution in the RD map. For the dense region, predict the non-target region where no target exists, and determine the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region, and perform target detection on the unit to be detected in the dense region; For the sparse region, the size information of the target corresponding to the unit to be detected is predicted, and the protection unit of the unit to be detected is determined based on the size information; A reference unit is determined based on the protection unit, and a noise power estimate of the unit to be detected in the sparse region is determined based on the reference unit. Target detection is then performed on the unit to be detected in the sparse region. The process of predicting the size information of the target corresponding to the unit to be detected, and determining the protection unit of the unit to be detected based on the size information, includes: Predict the type of the target corresponding to the unit to be detected, and determine the size information corresponding to the type; Based on the size information and the radar's range resolution, the number of protection units in the range dimension is determined; The number of protection units in the Doppler dimension is determined based on the preset acceleration of the target motion, the time interval between two adjacent RD images, and the radar velocity resolution. Based on the number of protection units in the distance dimension and the number of protection units in the Doppler dimension, protection units are determined in the neighborhood units of the unit to be detected.

2. The method according to claim 1, characterized in that, Determining the noise power estimate of the unit to be detected in the dense region based on at least one sub-region in the non-target region includes: Predict the distance range of the target distribution in the dense region, and determine the candidate region corresponding to the distance range in the non-target region; Select at least one sub-region from the candidate regions; The average energy value of each sub-region is determined, and the minimum value among the average energy values ​​is used as the noise power estimate of the unit to be detected in the dense region.

3. The method according to claim 1, characterized in that, Based on the size information and the radar's range resolution, the number of protection units in the range dimension is determined, including: Based on the ratio of the size information to the distance resolution, the number of distance units occupied by the target is predicted; wherein, the distance unit is the smallest unit in the distance dimension; The number of protection units in the distance dimension is determined based on the number of main lobe widening distance units after Fourier windowing processing in the distance dimension, and the number of distance units occupied by the predicted target.

4. The method according to claim 1, characterized in that, Based on the pre-set target acceleration, the time interval between two adjacent RD maps, and the radar's velocity resolution, the number of protection units in the Doppler dimension is determined, including: Based on the preset acceleration of the target motion and the time interval between two adjacent RD images, predict the change in the target's velocity. The number of Doppler cells occupied by the target is predicted based on the ratio of the velocity change to the velocity resolution. The number of protection cells in the Doppler dimension is determined based on the number of main lobe-widened Doppler cells after Fourier windowing in the Doppler dimension, and the number of Doppler cells occupied by the predicted target.

5. The method according to claim 1, characterized in that, The method further includes: Clustering is performed on the target detection results of the unit to be detected to obtain the targets in the RD diagram; Track the target in the RD map and determine the predicted position of the target in the RD map; Based on the predicted location, the number of tracks for each distance cell in the RD map is determined, and based on the number of tracks, the region type to which the distance cell belongs is determined, so as to predict the target distribution in the next frame of the RD map; wherein, the region type includes dense regions and sparse regions.

6. The method according to claim 5, characterized in that, The method further includes: The type of the target is determined based on the clustering results, and the type of the target corresponding to the unit to be detected in the next frame RD map is predicted based on the type of the target and the predicted position of the target.

7. A target detection device, characterized in that, The device includes: The region type prediction module is used to predict the dense and sparse regions of target distribution in the RD map based on the target detection results of the historical RD map if the range Doppler RD map is not the first frame RD map detected by the radar. The dense area target detection module is used to predict non-target areas where no target exists in the dense area, and determine the noise power estimate of the unit to be detected in the dense area based on at least one sub-region in the non-target area, and perform target detection on the unit to be detected in the dense area. The protection unit determination module is used to predict the size information of the target corresponding to the unit to be detected for the sparse region, and determine the protection unit of the unit to be detected based on the size information. A sparse region target detection module is used to determine a reference unit based on the protection unit, determine the noise power estimate of the unit to be detected in the sparse region based on the reference unit, and perform target detection on the unit to be detected in the sparse region. The protection unit determination module includes: A size information determination unit is used to predict the type of the target corresponding to the unit to be detected, and determine the size information corresponding to the type; The first protection unit quantity determination unit is used to determine the number of protection units in the range dimension based on the size information and the range resolution of the radar. The second protection unit is used to determine the number of protection units in the Doppler dimension based on the preset target motion acceleration, the time interval between two adjacent RD images, and the radar velocity resolution. A protection unit is used to determine protection units in the neighborhood units of the unit to be detected based on the number of protection units in the distance dimension and the number of protection units in the Doppler dimension.

8. A target detection electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the target detection method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the target detection method according to any one of claims 1-6.

Images