An aerial target single-photon laser three-dimensional histogram feature extraction method

Abstract

The application discloses an air target single-photon laser three-dimensional histogram feature extraction method, and the operation steps are composed of a time correlation noise reduction module, a histogram accumulation module and a histogram feature extraction module. The application can extract target features under the conditions that the scale of a detector pixel is limited, single imaging point cloud data is sparse and there is atmospheric turbulence interference, and the stability of the features is enhanced.

Description

Technical Field

[0001] This invention relates to a method for extracting three-dimensional histogram features of single-photon lasers from aerial targets, belonging to the field of radar imaging technology, and is applicable to the detection and identification of moving aerial targets using single-photon laser imaging. Background Technology

[0002] A single-photon detector (SPD) is a highly sensitive photodetector capable of responding to signals on the order of photons, and is fundamental to fields such as single-photon radar detection and quantum communication. Single-photon laser long-range 3D imaging detection and recognition technology utilizes a Geiger focal plane array detector to acquire a large amount of laser point cloud information in a single flash imaging process. It boasts advantages such as high imaging angular resolution, long operating range, simple structure, and fast imaging speed, enabling rapid acquisition of the long-range 3D contour features of targets, providing support for target recognition, and represents an important development direction for future lidar detection systems.

[0003] However, in existing target 3D imaging detection, the Geiger focal plane detector has a small pixel size, which has problems such as low target 3D contour resolution, sparse point cloud data acquired in a single acquisition, few extractable features, and susceptibility to atmospheric turbulence interference leading to deterioration of 3D imaging effect. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for extracting three-dimensional histogram features from single-photon laser beams of aerial targets. This method can effectively extract target features and achieve accurate target identification, thereby solving the aforementioned technical problems.

[0005] The present invention is achieved through the following technical solutions.

[0006] This invention provides a method for extracting three-dimensional histogram features of single-photon laser beams from airborne targets, comprising the following steps:

[0007] ① Data preprocessing: Receive raw data input from the detector, acquire preprocessed data, and complete data preprocessing;

[0008] ② Time-correlated noise reduction: Input preprocessed data, obtain valid data, and use time-correlated noise reduction algorithm to remove noise;

[0009] ③ Histogram accumulation: Input valid data to obtain histogram data, where the horizontal axis of the histogram data is the value T, and the vertical axis is the number of pixels with the value T;

[0010] ④ Histogram feature extraction: Input histogram data and obtain the target feature vector.

[0011] In step ①, the image data of images that have not received an optical signal is set to zero, while the image data of images that have received an optical signal remains unchanged.

[0012] In step ②, the valid data is the target three-dimensional imaging data without background noise.

[0013] The pixel value T≤T min or T≥T max Setting the data in a range to zero indicates that the data for this pixel is invalid, while the rest of the data remains unchanged. Here, the value T represents the time from when a certain pixel of the detector starts working until it receives a light signal.

[0014] In step ②, the processing method of the time-correlated noise reduction algorithm is as follows:

[0015] (1) Take an n×n operator from the imaging data;

[0016] (2) Subtract each non-zero data in the operator from the central data of the operator;

[0017] (3) The number of statistical differences whose absolute values ​​are less than the threshold V1, n1;

[0018] (4) When n1 is less than the threshold V2, the center data is determined to be noise data and the value of the pixel is set to zero. Otherwise, it is valid data and the original data value is retained.

[0019] Step ③ specifically consists of the following steps:

[0020] (3.1) Single-frame histogram statistics: Statistically analyze the histogram of data for each frame of the detector;

[0021] (3.2) Histogram low-pass filtering: Take the average of m values ​​before and after a certain digital value as the value at that moment;

[0022] (3.3) Histogram registration:

[0023] a. Calculate the centroid position of the histogram data. The calculation method is as follows:

[0024]

[0025] In the formula, T m Let n be the centroid of the histogram, t be a time point on the x-axis of the histogram, and n be the x-axis of the histogram. t Let N be the number of pixels at time t, and N be the total number of valid pixels.

[0026] b. Calculate the centroid deviation value ΔT of the histogram: ΔT = Ts - Tm, T s It is a fixed constant;

[0027] c. Histogram translation: Shift the histogram data along the x-axis by ΔT.

[0028] (3.4) Histogram accumulation: Add the histogram data of several adjacent frames after registration.

[0029] In step ④, the target contour features are extracted from the histogram accumulation results, and the target feature vector is output.

[0030] The target features include, but are not limited to, histogram-based features such as time width, number of peaks, and peak distance.

[0031] The beneficial effects of this invention are: it can effectively extract target features under conditions of limited detector pixel size, sparse single-image point cloud data, and atmospheric turbulence interference, thereby achieving accurate target identification and enhancing feature stability. Attached Figure Description

[0032] Figure 1 This is a flowchart of the present invention;

[0033] Figure 2 This is an example diagram of histogram statistics data from the present invention;

[0034] Figure 3 This is an example diagram of histogram feature extraction according to the present invention. Detailed Implementation

[0035] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0036] It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, it should be pointed out that unless otherwise specified, the structures, connections, positional relationships, power source relationships, etc., involved in this invention are all things that those skilled in the art can discover without creative effort based on existing technology.

[0037] like Figure 1 As shown, a method for extracting three-dimensional histogram features from a single-photon laser beam of an airborne target includes the following steps:

[0038] ① Data preprocessing: Receive raw data input from the detector, acquire preprocessed data, and complete data preprocessing;

[0039] ② Time-correlated noise reduction: Input preprocessed data, obtain valid data, and use time-correlated noise reduction algorithm to remove noise;

[0040] ③ Histogram accumulation: Input valid data to obtain histogram data, where the horizontal axis of the histogram data is the value T, and the vertical axis is the number of pixels with the value T;

[0041] ④ Histogram feature extraction: Input histogram data and obtain the target feature vector.

[0042] In step ①, the image data of images that have not received an optical signal is set to zero, while the image data of images that have received an optical signal remains unchanged.

[0043] In step ②, the valid data is the target three-dimensional imaging data without background noise.

[0044] The pixel value T≤T min or T≥T max Setting the data in a range to zero indicates that the data for this pixel is invalid, while the rest of the data remains unchanged. Here, the value T represents the time from when a certain pixel of the detector starts working until it receives a light signal.

[0045] In step ②, the processing method of the time-correlated noise reduction algorithm is as follows:

[0046] (1) Take an n×n operator from the imaging data;

[0047] (2) Subtract each non-zero data in the operator from the central data of the operator;

[0048] (3) The number of statistical differences whose absolute values ​​are less than the threshold V1, n1;

[0049] (4) When n1 is less than the threshold V2, the center data is determined to be noise data and the value of the pixel is set to zero. Otherwise, it is valid data and the original data value is retained.

[0050] Step ③ specifically consists of the following steps:

[0051] (3.1) Single-frame histogram statistics: The histogram of each frame of data from the detector is statistically analyzed, where the horizontal axis represents the value T, and the vertical axis represents the number of pixels with value T. For example... Figure 2 As shown;

[0052] (3.2) Histogram low-pass filtering: Take the average of m values ​​before and after a certain digital value as the value at that moment;

[0053] (3.3) Histogram registration:

[0054] a. Calculate the centroid position of the histogram data. The calculation method is as follows:

[0055]

[0056] In the formula, T m Let n be the centroid of the histogram, t be a time point on the x-axis of the histogram, and n be the x-axis of the histogram. t Let N be the number of pixels at time t (vertical axis), and N be the total number of valid pixels.

[0057] b. Calculate the centroid deviation value ΔT of the histogram: ΔT = Ts - Tm, T s It is a fixed constant;

[0058] c. Histogram translation: Shift the histogram data along the x-axis by ΔT.

[0059] (3.4) Histogram accumulation: The histogram data of several adjacent frames after registration are added together to enhance their histogram features.

[0060] In step ④, the target contour features are extracted from the histogram accumulation results, and the target feature vector is output.

[0061] Target features include, but are not limited to, histogram-based features such as time width, number of peaks, and peak distance, etc. Figure 3 As shown, features are extracted from the histogram:

[0062] (1) Time width d1;

[0063] (2) Number of peak values;

[0064] (3) The peak distances d2, d3, and d4 between peaks;

[0065] (4) Other features related to histograms.

[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for extracting three-dimensional histogram features from single-photon laser beams of aerial targets, characterized in that: Includes the following steps: ① Data preprocessing: Receive raw data input from the detector, acquire preprocessed data, and complete data preprocessing; ② Time-correlated noise reduction: Input preprocessed data, obtain valid data, and use time-correlated noise reduction algorithm to remove noise; ③ Histogram accumulation: Input valid data to obtain histogram data, where the horizontal axis of the histogram data is the value T, and the vertical axis is the number of pixels with the value T; where the value T is the time from when a certain pixel of the detector starts working to when it receives the light signal. ④ Histogram feature extraction: Input histogram data and obtain the target feature vector; In step ②, the processing method of the time-correlated noise reduction algorithm is as follows: (1) Take an n×n operator from the imaging data; (2) Subtract each non-zero data in the operator from the central data of the operator; (3) The number of statistical differences with an absolute value less than the threshold V1, n1; (4) When n1 is less than the threshold V2, the center data is determined to be noise data and the value of the pixel is set to zero. Otherwise, it is valid data and the original data value is retained.

2. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 1, characterized in that: In step ①, the image data of images that have not received an optical signal is set to zero, while the image data of images that have received an optical signal remains unchanged.

3. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 1, characterized in that: In step ②, the valid data is the target three-dimensional imaging data without background noise.

4. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 2, characterized in that: Pixel value T≤ or T≥ Setting the data in a range to zero indicates that the data for that pixel is invalid, while the rest of the data remains unchanged.

5. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 1, characterized in that: Step ③ specifically consists of the following steps: (3.1) Single-frame histogram statistics: Statistically analyze the histogram of data for each frame of the detector; (3.2) Histogram low-pass filtering: Take the average of m values ​​before and after a certain numerical value as the value of that numerical value T; (3.3) Histogram registration: a. Calculate the centroid position of the histogram data. The calculation method is as follows: In the formula, Let the centroid of the histogram be... This represents a time point on the x-axis of the histogram. Let N be the number of pixels at time t, and N be the total number of valid pixels. b. Calculate the centroid deviation of the histogram. : , It is a fixed constant; c. Histogram translation: Translate the histogram data along the x-axis. ; (3.4) Histogram accumulation: The histogram data of several adjacent frames after registration are added together.

6. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 1, characterized in that: In step ④, the target contour features are extracted from the histogram accumulation results, and the target feature vector is output.

7. The method for extracting three-dimensional histogram features of a single-photon laser beam from an aerial target as described in claim 6, characterized in that: The target features include, but are not limited to, histogram-based features such as time width, number of peaks, and peak distance.

Images