A target detection method based on CFAR and one-dimensional range profile

Abstract

The present invention relates to a target detection method based on CFAR and one-dimensional range image, comprising the following steps: acquiring ADC data and performing a range dimension FFT operation to extract a one-dimensional range image with zero Doppler; performing a speed dimension FFT and CFAR detection algorithm operation on a result of the range dimension FFT; performing an angle estimation operation on points after CFAR and points of the one-dimensional range image above a threshold, and outputting point cloud data; finally, judging whether the point cloud data is located in a detection area, if the point cloud data is located in the detection area, it is considered that there is a target in the detection area, and a gate rod is controlled not to fall; otherwise, it is considered that there is no target in the detection area, and the gate rod is controlled to fall; the method provided by the present invention can accurately and stably detect targets in the area, especially obliquely parked vehicles, to prevent the gate radar from hitting the vehicle.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing technology, and in particular to a target detection method based on CFAR and one-dimensional range image. Background Art

[0002] With the development of technology, parking lots are also showing a trend of intelligence, so barrier anti-smashing radar came into being; by setting the barrier drop bar to control the gate machine, it can avoid the situation where the bar falls and smashes the vehicle before the vehicle leaves the detection area; when the diagonal vehicle just drives into the detection area, the radar can detect the vehicle target normally, however, when the vehicle drives forward and stops diagonally in the detection area, the large area of ​​metal on the side of the vehicle refracts most of the useful signals, and the echo signal received by the radar is weak, resulting in the radar being unable to accurately detect the vehicle target and being judged as noise by the CFAR detection algorithm, causing the radar to miss the diagonal vehicle and smash the car.

[0003] In order to enhance the detection capability of weak targets, CN202110581785.6 discloses a method for detecting targets based on one-dimensional range image recording background comparison, and its main steps include: the radar collects a one-dimensional range image sequence when there is no target and calculates the variance; collects and processes the signal to obtain a one-dimensional range image sequence of the current frame; the radar executes the CFAR detection algorithm for angle estimation and outputs point cloud data; calculates the variance of the one-dimensional range image of the current frame, and performs correlation calculation on the one-dimensional range image sequence of the current frame and the stored one-dimensional range image sequence; compares the correlation result with the threshold, and when the correlation is greater than the threshold and the variance of the one-dimensional range image sequence of the current frame is less than the variance of the one-dimensional range image sequence of the empty scene, and the radar does not detect the point cloud data, it is determined that there is no target; otherwise, it is determined that there is a target; this method can remove environmental interference, retain effective target information and detect vehicles entering at a large angle; however, the method provided by this patent needs to be combined with an empty scene for judgment, and does not estimate the angle of the points on the one-dimensional range image, nor does it increase the output of the point cloud; in addition, the basic idea of ​​the CFAR detection algorithm is to compare the detection unit with the reference unit. If the energy of the detection unit is higher than that of the reference unit, it is considered that there is a target in the detection unit, otherwise it is considered that there is no target in the detection unit; and the energy values ​​on the sides of the diagonally parked vehicles are almost the same, so there are diagonally parked vehicles that cannot be detected; therefore, the accuracy and stability of this target detection method are not high. Summary of the invention

[0004] In view of this, the present invention provides a target detection method based on CFAR and one-dimensional range image, which effectively improves the accuracy and stability of target detection of the prior art millimeter-wave barrier anti-smashing radar, especially the identification of vehicles entering at a large angle and vehicles parked at an angle.

[0005] To achieve the above purpose, a target detection method based on CFAR and one-dimensional range image includes the following steps:

[0006] S1, obtain the original radar ADC data, perform range-dimensional FFT processing and extract a one-dimensional range image sequence with zero Doppler;

[0007] S2, performing speed dimension FFT and CFAR detection algorithm processing on the result of distance dimension FFT processing;

[0008] S3, points processed by CFAR detection algorithm and above the threshold The one-dimensional distance image of the point is used to perform angle estimation and output point cloud data ;

[0009] S4: Determine whether the point cloud data is within the detection area.

[0010] If point cloud data exists satisfy:

[0011] ;

[0012] If the target is in the detection area, it is considered that there is a target, and the gate rod is controlled not to fall. Otherwise, it is considered that there is no target in the detection area, and the gate rod is controlled to fall, realizing the anti-smashing function of the gate radar.

[0013] in, , K represents the number of point clouds, , and Respectively represent the abscissa, ordinate and energy value of the kth point, represents the absolute value function, Indicates the threshold value of each segment after segmentation.

[0014] Preferably, a millimeter wave radar waveform is configured, electromagnetic waves are received and sent to collect raw echo data, ADC data is collected on the raw echo data, and then a range-dimensional FFT operation is performed to extract a zero-Doppler one-dimensional range image sequence.

[0015] Preferably, the distance dimension FFT and speed dimension FFT processing are used to obtain the distance velocity heat map, and then the distance dimension CFAR and speed dimension CFAR detection are performed to obtain the point ,in , , P and Q represent the number of points processed by distance dimension CFAR and velocity dimension CFAR respectively.

[0016] Preferably, the point And select the point on the zero Doppler one-dimensional range image Angle estimation is performed, where , M represents the number of zero Doppler one-dimensional range image sequence points; output point cloud data .

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] The method provided by the present invention can improve the accuracy and stability of the targets in the detection area, especially the vehicles parked at an angle and the vehicles entering at a large angle, and enhance the working performance of the barrier radar in a complex environment;

[0019] The method provided by the present invention performs angle estimation after selecting points on the range image, selects point cloud targets located in the detection area, and greatly reduces the phenomenon of gate radar working errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of the algorithm of the present invention;

[0021] Figure 2 This is a schematic diagram of the vehicle being parked diagonally according to the present invention;

[0022] Figure 3 This is a distance image of an obliquely parked vehicle according to the present invention;

[0023] Figure 4 This is the principle diagram of the CFAR-CASO detection algorithm;

[0024] Figure 5 This is a point cloud image of an obliquely parked vehicle based on a range image according to the present invention. DETAILED DESCRIPTION

[0025] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0026] Existing technology detection targets such as Figure 2 As shown, Figure 2 This is a schematic diagram of a vehicle parked diagonally in the detection area. The near-end longitudinal non-detection distance is , the longitudinal detection distance is , the horizontal detection distance range is ; The dotted line is the radar detection boundary, and the detection boundary is the detection area; if there is a point cloud satisfy and , then the point cloud p is considered to be within the detection area; since the side of the diagonally parked vehicle is a large flat metal surface, it refracts most of the useful signals, and the echo signal received by the radar is weak, resulting in no point cloud target output in the detection area, which can easily cause the radar to miss the diagonally parked vehicle and cause the vehicle to be hit.

[0027] In order to effectively improve the recognition accuracy of the barrier radar for vehicles entering at a large angle and vehicles parked at an angle, and reduce the risk of the barrier radar hitting the target by mistake, the present invention provides a target detection method based on CFAR and one-dimensional range image, comprising the following steps:

[0028] S1, obtain the original radar ADC data, perform range-dimensional FFT processing and extract a one-dimensional range image sequence with zero Doppler;

[0029] Configure the millimeter-wave radar waveform, receive and send electromagnetic waves to collect raw echo data, perform ADC data acquisition on the raw echo data, and then perform one-dimensional fast Fourier transform processing, that is, range-dimensional FFT operation. The data after range-dimensional FFT processing includes range-dimensional FFT data of each different Doppler dimension, and zero-Doppler one-dimensional range image data is extracted from these Doppler-dimensional range-dimensional FFT data.

[0030] S2, performing speed dimension FFT and CFAR detection algorithm processing on the result of distance dimension FFT processing;

[0031] The distance-speed heat map is obtained by processing the distance-dimensional FFT and speed-dimensional FFT, and then the distance-dimensional CFAR and speed-dimensional CFAR detection are performed to obtain the point ,in , , P and Q represent the number of points processed by distance dimension CFAR and velocity dimension CFAR respectively;

[0032] There are many one-dimensional CFARs available for distance dimension CFAR and velocity dimension CFAR, such as CFAR-CA, CFAR-CAGO and CFAR-CASO. The schematic diagram of CFAR-CASO is shown in the figure. Figure 4 As shown, if the energy value corresponding to the detection unit D is satisfy , it is considered that there is a target in the detection unit, and its distance dimension index and speed dimension index are output. , otherwise it is considered that there is no target in the detection unit; distance dimension CFAR and speed dimension CFAR can also choose two-dimensional CFAR.

[0033] S3, points processed by CFAR detection algorithm and above the threshold The one-dimensional distance image of the point is used to perform angle estimation and output point cloud data ;

[0034] The one-dimensional range image of a vehicle parked at an angle with zero Doppler is as follows: Figure 3 As shown in the figure, since the radar energy will decay as the distance increases or decreases, different thresholds are set at different distances. ( ), where N is the number of distance segments; if it is above the threshold , select the points on these Doppler one-dimensional range images ,in , M represents the number of zero Doppler one-dimensional range image sequence points, that is, Figure 3 The point marked in red; the point And select the point on the zero Doppler one-dimensional range image Perform angle estimation. The available angle estimation algorithms include angle FFT and DBF, and output point cloud data .

[0035] S4: Determine whether the point cloud data is within the detection area.

[0036] If point cloud data exists satisfy:

[0037] ;

[0038] If the target is in the detection area, it is considered that there is a target, and the gate rod is controlled not to fall; otherwise, it is considered that there is no target in the detection area, and the gate rod is controlled to fall;

[0039] in, , K represents the number of point clouds, , and Respectively represent the abscissa, ordinate and energy value of the kth point, represents the absolute value function, Represents the threshold value of each segment after segmentation, that is, Divided into N segments, the threshold of each segment is ;like Figure 5 As shown in the figure, when a vehicle is parked obliquely in the detection area, after the angle estimation of the selected points on the zero-Doppler one-dimensional range image, the output point cloud is located in the detection area (the points marked in red), then it can be detected that there is a vehicle parked in the detection area, and the pole will not fall and hit the car; Figure 5 In the figure, the black dotted line indicates the horizontal detection range. The green dotted line indicates the horizontal detection range is within Within.

[0040] After many experiments, it was found that the smaller the lateral detection range, the more obvious the target detection method after integrating CFAR and one-dimensional range image is. The selected points on the zero-Doppler one-dimensional range image are more helpful in detecting the point cloud and preventing the car from being smashed.

[0041] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A target detection method based on CFAR and one-dimensional range image, characterized in that: The following steps are involved: S1, obtain the original radar ADC data, perform range-dimensional FFT processing and extract a one-dimensional range image sequence with zero Doppler; S2, performing speed dimension FFT and CFAR detection algorithm processing on the result of distance dimension FFT processing; S3, points processed by CFAR detection algorithm and above the threshold The one-dimensional distance image of the point is used to perform angle estimation and output point cloud data ; The distance-speed heat map is obtained by processing the distance-dimensional FFT and speed-dimensional FFT, and then the distance-dimensional CFAR and speed-dimensional CFAR detection are performed to obtain the point ,in , , P and Q represent the number of points processed by distance dimension CFAR and velocity dimension CFAR respectively. And select the point on the zero Doppler one-dimensional range image Angle estimation is performed, where , M represents the number of zero Doppler one-dimensional range image sequence points; output point cloud data ; S4: Determine whether the point cloud data is within the detection area. If point cloud data exists satisfy: ； If the target is in the detection area, it is considered that there is a target, and the gate rod is controlled not to fall; otherwise, it is considered that there is no target in the detection area, and the gate rod is controlled to fall; in, , K represents the number of point clouds, , and Respectively represent the abscissa, ordinate and energy value of the kth point, represents the absolute value function, Indicates the threshold value of each segment after segmentation.

2. The target detection method based on CFAR and one-dimensional range profile according to claim 1, characterized in that: Configure the millimeter-wave radar waveform, receive and send electromagnetic waves to collect raw echo data, perform ADC data acquisition on the raw echo data, perform range-dimensional FFT operation, and extract a zero-Doppler one-dimensional range image sequence.

Images