CN102063615A

CN102063615A - Beacon light optimizing and recognizing denoising method based on spot noise distribution topological characteristic

Info

Publication number: CN102063615A
Application number: CN 201010611190
Authority: CN
Inventors: 王强; 马晶; 谭立英; 于思源; 韩琦琦; 付森; 宋义伟
Original assignee: Harbin Institute of Technology Shenzhen
Current assignee: Harbin Institute of Technology Shenzhen
Priority date: 2010-12-29
Filing date: 2010-12-29
Publication date: 2011-05-18
Anticipated expiration: 2030-12-29
Also published as: CN102063615B

Abstract

A beacon light optimal recognition noise reduction method based on point noise distribution topology characteristics relates to a beacon light optimal recognition noise reduction method. It solves the problem of poor real-time performance due to the large amount of calculation in the existing beacon light optimization identification and noise reduction methods. The method: use the CCD camera to perform threshold segmentation processing on the beacon light image at the current moment to obtain the spot centroid distribution image; calculate the semicircle area discriminant ΔS ₃ of the eigenvalue vector of the centroid distribution of each spot, and the arc of the eigenvalue vector Long discriminant ΔL ₃ and centroid coordinate change rate discriminant ΔC; and judge. In this way, continuous tracking, identification and noise reduction of background light and beacon light are realized. The invention is suitable for beacon light optimization identification and noise reduction.

Description

A Beacon Light Optimization Recognition and Noise Reduction Method Based on Topological Characteristics of Point Noise Distribution

技术领域technical field

本发明涉及一种信标光优化识别降噪方法。The invention relates to a beacon light optimization identification noise reduction method.

背景技术Background technique

卫星激光通信系统光电图像识别需要在空间恒星背景下进行，恒星在光电探测器上形成点像，探测器需要在诸多点像中甄别实际需要跟踪的光束点像，这些光束点像即为点噪声。这些点噪声会使得信标光的识别变得较困难，严重时会使得卫星光通信系统的指向偏差，影响链路的建立和稳定。消除点噪声影响的通常采用的方法是查询星表，具体过程是，将点噪声与恒星背景进行匹配，判断其是否为恒星背景光，这种方法的缺点是计算量比较大，实时性较差，不便于在卫星光通信中应用。The photoelectric image recognition of the satellite laser communication system needs to be carried out in the background of space stars. The stars form point images on the photodetector. The detector needs to identify the actual beam point images that need to be tracked among many point images. These beam point images are point noises. . These point noises will make it difficult to identify the beacon light, and in severe cases, it will cause the pointing deviation of the satellite optical communication system, affecting the establishment and stability of the link. The usual method to eliminate the influence of point noise is to query the star catalog. The specific process is to match the point noise with the star background to determine whether it is the star background light. The disadvantage of this method is that the calculation amount is relatively large and the real-time performance is poor. , it is not convenient to apply in satellite optical communication.

发明内容Contents of the invention

本发明是为了解决现有的信标光优化识别降噪方法由于计算量大导致实时性较差的问题，从而提供一种基于点噪声分布拓扑特性的信标光优化识别降噪方法。The present invention aims to solve the problem of poor real-time performance due to large amount of calculation in the existing beacon light optimization identification and noise reduction method, thereby providing a beacon light optimization identification noise reduction method based on point noise distribution topology characteristics.

一种基于点噪声分布拓扑特性的信标光优化识别降噪方法，它的方法是：A beacon light optimization identification noise reduction method based on point noise distribution topological characteristics, its method is:

采用CCD相机从初始时刻开始连续采集多幅图像，CCD相机的采样周期为n，CCD相机的采样率为f_CCD；Adopt the CCD camera to start collecting multiple images continuously from the initial moment, the sampling period of the CCD camera is n, and the sampling rate of the CCD camera is _fCCD ;

将初始时刻的下一时刻作为当前时刻，将该时刻所采集到的图像作为当前时刻的图像；Taking the moment next to the initial moment as the current moment, and the image collected at this moment as the image of the current moment;

步骤一、将当前时刻的图像进行阈值分割处理，获得光斑质心分布图像；Step 1. Perform threshold segmentation processing on the image at the current moment to obtain a spot centroid distribution image;

步骤二、根据步骤一获得的光斑质心分布图像，对于其中每一个光斑，均通过公式：Step 2. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

$\frac{{S S}_{33}^{' '}}{{S S}_{33}} = = 11 + + Δ Δ {S S}_{33}$

计算每个光斑的质心分布的特征值向量的半圆面积判别量ΔS₃；式中：S′₃和S₃分别为当前时刻的前一时刻和后一时刻下的CCD相机所采集图像的点噪声分布的特征值向量的半圆面积参量；Calculate the semicircle area discriminant ΔS ₃ of the eigenvalue vector of the centroid distribution of each spot; where: S′ ₃ and S ₃ are the point noise of the image collected by the CCD camera at the previous moment and the next moment of the current moment respectively the semicircle area parameter of the eigenvalue vector of the distribution;

步骤三、根据步骤一获得的光斑质心分布图像，对于其中每一个光斑，均通过公式：Step 3. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

$\frac{{L L}_{33}^{' '}}{{L L}_{33}} = = 11 + + Δ Δ {L L}_{33}$

计算每个光斑的质心分布的特征值向量的弧长判别量ΔL₃；式中：L′₃和L₃分别为当前时刻的前一时刻和后一时刻下的CCD相机所采集图像中点噪声分布的特征值向量的弧长参量；Calculate the arc length discriminant ΔL ₃ of the eigenvalue vector of the centroid distribution of each spot; where: L′ ₃ and L ₃ are the point noise in the image collected by the CCD camera at the previous moment and the next moment of the current moment respectively the arc length parameter of the eigenvalue vector of the distribution;

步骤四、根据步骤一获得的光斑质心分布图像，对于其中每一个光斑，均通过公式：Step 4. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

$\frac{\sqrt{{(({x x}_{i i 11} - - {x x}_{i i 22}))}^{22} + + {(({y the y}_{i i 11} - - {y the y}_{i i 22}))}^{22}}}{n no \cdot \cdot \frac{11}{{f f}_{CCD CCD}}} = = 11 + + ΔC ΔC$

计算该光斑的质心坐标变化率判别量ΔC；式中：(x_i1，y_i1)为第i个光斑初始时刻的质心坐标，(x_i2，y_i2)为第n个采样周期中获得的光斑质心坐标，K为CCD相机视域内的光斑个数，i、n均为正整数；Calculate the discriminant ΔC of the centroid coordinate change rate of the spot; where (x _i1 , y _i1 ) is the centroid coordinate of the i-th spot at the initial moment, (x _i2 , y _i2 ) is the spot obtained in the n-th sampling period Centroid coordinates, K is the number of spots in the field of view of the CCD camera, i and n are both positive integers;

步骤五、遍历光斑质心分布图像中的每个光斑，进行信标光识别，对每个光斑的识别过程为：判断该光斑的质心分布的特征值向量的半圆面积判别量ΔS₃、计算光斑质心分布的特征值向量的弧长判别量ΔL₃是否均小于5个像素，如果判断结果为是，则执行步骤五一；如果判断结果为否，则执行步骤五二；Step 5. Traversing each spot in the spot centroid distribution image to identify the beacon light. The identification process for each spot is: judging the semicircle area discriminant ΔS ₃ of the eigenvalue vector of the centroid distribution of the spot, and calculating the spot centroid Whether the arc length discrimination quantity ΔL ₃ of the distributed eigenvalue vector is less than 5 pixels, if the judgment result is yes, then execute step 51; if the judgment result is no, then execute step 52;

步骤五一、判断该光斑的质心坐标变化率判别量ΔC是否小于五个像素/每秒钟，如果判断结果为是，则该光斑为恒星背景光斑，当遍历所有的光斑均为恒星背景光斑时，表示当前时刻的图像中没有信标光光斑，执行步骤六；如果判断结果为否，则执行步骤五二；Step 51. Determine whether the discriminant value ΔC of the centroid coordinate change rate of the spot is less than five pixels per second. If the judgment result is yes, the spot is a star background spot. When all the spots traversed are stellar background spots , indicating that there is no beacon light spot in the image at the current moment, go to step 6; if the judgment result is no, go to step 52;

步骤五二、将所有光斑的质心坐标变化率判别量ΔC进行比较，其中质心坐标变化率判别量ΔC最大者所对应的光斑为当前时刻的信标光光斑，实现信标光识别降噪，并执行步骤六；Step 52: Compare the discriminant amount ΔC of the centroid coordinate change rate of all light spots, and the spot corresponding to the largest discriminant amount ΔC of the centroid coordinate change rate is the beacon light spot at the current moment, so as to realize beacon light recognition and noise reduction, and Execute step six;

步骤六、将下一个时刻对应的图像作为当前时刻的图像，返回步骤一继续进行信标光识别。Step 6. Use the image corresponding to the next moment as the image at the current moment, and return to step 1 to continue beacon light recognition.

有益效果：本发明提出一种基于点噪声分布拓扑特性的信标光优化识别降噪方法，充分利用CCD图像点噪声分布拓扑结构的特征值，实现了对信标光的准确识别，由于此种方法的计算量很小，所以具有良好的实时性。Beneficial effects: the present invention proposes a beacon light optimization identification noise reduction method based on point noise distribution topological characteristics, fully utilizes the characteristic value of the CCD image point noise distribution topology structure, and realizes accurate identification of beacon light. The calculation amount of the method is very small, so it has good real-time performance.

附图说明Description of drawings

图1是具体实施方式一中点噪声分布特征值计算示意图。Fig. 1 is a schematic diagram of the calculation of the characteristic value of the midpoint noise distribution in the specific embodiment one.

具体实施方式Detailed ways

具体实施方式一、一种基于点噪声分布拓扑特性的信标光优化识别降噪方法，它的方法是：Specific embodiments 1. A beacon light optimization identification and noise reduction method based on point noise distribution topology characteristics, its method is:

$\frac{{S S}_{33}^{' '}}{{S S}_{33}} = = 11 + + Δ Δ {S S}_{33}$

$\frac{{L L}_{33}^{' '}}{{L L}_{33}} = = 11 + + Δ Δ {L L}_{33}$

$\frac{\sqrt{{(({x x}_{i i 11} - - {x x}_{i i 22}))}^{22} + + {(({y the y}_{i i 11} - - {y the y}_{i i 22}))}^{22}}}{n no \cdot &Center Dot; \frac{11}{{f f}_{CCD CCD}}} = = 11 + + ΔC ΔC$

步骤五一、判断该光斑的质心坐标变化率判别量ΔC是否小于五个像素/每秒钟，如果判断结果为是，则该光斑为恒星背景光斑，当遍历所有的光斑均为恒星背景光斑时，表示当前时刻的图像中没有信标光光斑，执行步骤六；如果判断结果为否，则执行步骤五二；Step 51. Determine whether the center of mass coordinate change rate discrimination value ΔC of the spot is less than five pixels per second. If the judgment result is yes, the spot is a star background spot. , indicating that there is no beacon light spot in the image at the current moment, go to step 6; if the judgment result is no, go to step 52;

重复上述步骤，实现对信标光的跟踪。Repeat the above steps to realize the tracking of the beacon light.

工作原理：首先定义遍历CCD图像全部光斑的半圆弧封闭曲线长度和各个半圆面积总和为光斑分布的特征值向量，前者称为弧长参量，后者称为半圆参量，显然，这个特征值向量是一个二维向量。在图1中所示，各个半圆弧段长度之和为P₀₁+P₂₀+P₁₂，如果S₁是由光斑0和光斑1所成直线段与连接这两个光斑的半圆弧线P₀₁围成的半圆盘面积，其余记号的理解同理。那么，全部半圆面积总和为S₁+S₂+S₃。此时，点噪声分布的特征值向量就是二维向量(P₀₁+P₂₀+P₁₂，S₁+S₂+S₃)。如果光斑1在光斑0和光斑2所形成的直线段上运动，点噪声分布的特征值向量(P₀₁+P₂₀+P₁₂，S₁+S₂+S₃)将会发生变化，即特征值向量能够体现点噪声分布的变化。Working principle: First, define the length of the closed curve of the semicircle that traverses all the spots of the CCD image and the sum of the areas of each semicircle as the eigenvalue vector of the spot distribution. The former is called the arc length parameter, and the latter is called the semicircle parameter. Obviously, this eigenvalue vector is a two-dimensional vector. As shown in Figure 1, the sum of the lengths of each semicircle segment is P ₀₁ + P ₂₀ + P ₁₂ , if S ₁ is a straight line segment formed by spot 0 and spot 1 and the semicircle arc line P connecting these two spots The area of the semi-disc surrounded by ₀₁ , the understanding of other symbols is the same. Then, the sum of the areas of all semicircles is S ₁ +S ₂ +S ₃ . At this time, the eigenvalue vector of the point noise distribution is a two-dimensional vector (P ₀₁ +P ₂₀ +P ₁₂ , S ₁ +S ₂ +S ₃ ). If spot 1 moves on the straight line segment formed by spot 0 and spot 2, the eigenvalue vector (P ₀₁ +P ₂₀ +P ₁₂ , S ₁ +S ₂ +S ₃ ) of the point noise distribution will change, that is, the characteristic The value vector can reflect the change of point noise distribution.

为了给出利用图像点噪声分布的特征值的信标光捕获识别方法，先从连续两帧采样图像点噪声分布特征值构造捕获识别信标光的两个判别量。In order to give a beacon light capture and recognition method using the eigenvalues of image point noise distribution, two discriminants for capturing and identifying beacon light are constructed from the eigenvalues of image point noise distribution in two consecutive frames.

令：make:

$\frac{{S S}_{33}^{' '}}{{S S}_{33}} = = 11 + + Δ Δ {S S}_{33}$

其中，S′₃和S₃分别为前后两时刻CCD采集图像所得点噪声分布的特征值向量的半圆面积参量，ΔS₃是半圆面积判别量，它主要由系统的综合误差、点噪声和信标光的移动共同决定。令：Among them, S′ ₃ and S ₃ are the semicircle area parameters of the eigenvalue vectors of the point noise distribution _obtained from the images collected by the CCD at two moments before and after, respectively. The move is jointly decided. make:

$\frac{{L L}_{33}^{' '}}{{L L}_{33}} = = 11 + + Δ Δ {L L}_{33}$

其中，L′₃和L₃分别为前后两时刻CCD采集图像所得点噪声分布的特征值向量的弧长参量，ΔL₃是弧长判别量，它主要由系统的综合误差、点噪声和信标光的移动共同决定。Among them, L′ ₃ and L ₃ are the arc length parameters of the eigenvalue vectors of the point noise distribution obtained from the images collected by the CCD at two moments before and after, respectively, and ΔL ₃ is the discriminant quantity of the arc length, which is mainly determined by the comprehensive error of the system, point noise and beacon light. The move is jointly decided.

$\frac{\sqrt{{(({x x}_{i i 11} - - {x x}_{i i 22}))}^{22} + + {(({y the y}_{i i 11} - - {y the y}_{i i 22}))}^{22}}}{n no \cdot \cdot \frac{11}{{f f}_{CCD CCD}}} = = 11 + + ΔC ΔC ((i i = = 1,2,3 1,2,3,, . . . . . . . . . . . .,, K K))$

其中，(x_i1，y_i1)为第i个光斑初始时刻的质心坐标，n为CCD的采样周期，f_CCD为CCD的采样率，(x_i2，y_i2)为第n个采样周期时的质心坐标，ΔC是利用点噪声分布中质心坐标缓慢变化特性进行信标光捕获识别的坐标变化率判别量，它是一个补充判别量，主要由系统综合误差所决定，K为CCD视域内的光斑个数。Among them, (x _i1 , y _i1 ) is the centroid coordinates of the i-th spot at the initial moment, n is the sampling period of the CCD, f _CCD is the sampling rate of the CCD, (x _i2 , y _i2 ) is the Centroid coordinates, ΔC is the discriminant quantity of the coordinate change rate of the beacon light capture and recognition by using the slow change characteristics of the centroid coordinates in the point noise distribution, it is a supplementary discriminant quantity, mainly determined by the comprehensive error of the system, K is the light spot in the CCD field of view number.

本实施方式中选用台湾敏通公司生产的MTV-1801的CCD相机，主要参数如下：像元数795(H)×596(V)；光谱响应范围400nm～1100nm；分辨率600TVL；探测灵敏度0.021lx；信噪比大于46dB；工作温度为-10℃～50℃。并且采用基于1394协议的视频数据采集卡，将图像信息输入计算机。Select the CCD camera of MTV-1801 produced by Taiwan Mintong Company for use in this embodiment, the main parameters are as follows: number of pixels 795 (H) × 596 (V); spectral response range 400nm～1100nm; resolution 600TVL; detection sensitivity 0.021lx ; The signal-to-noise ratio is greater than 46dB; the working temperature is -10℃～50℃. And the video data acquisition card based on the 1394 protocol is used to input the image information into the computer.

Claims

1. A beacon light optimization identification noise reduction method based on point noise distribution topological characteristics, is characterized in that: its method is:

Adopt the CCD camera to start collecting multiple images continuously from the initial moment, the sampling period of the CCD camera is n, and the sampling rate of the CCD camera is _fCCD ;

Taking the moment next to the initial moment as the current moment, and the image collected at this moment as the image of the current moment;

Step 1. Perform threshold segmentation processing on the image at the current moment to obtain a spot centroid distribution image;

Step 2. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

\frac{{S S}_{33}^{' '}}{{S S}_{33}} = = 11 + + Δ Δ {S S}_{33}

Calculate the semicircle area discriminant ΔS ₃ of the eigenvalue vector of the centroid distribution of each spot; where: S′ ₃ and S ₃ are the point noise of the image collected by the CCD camera at the previous moment and the next moment of the current moment respectively the semicircle area parameter of the eigenvalue vector of the distribution;

Step 3. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

\frac{{L L}_{33}^{' '}}{{L L}_{33}} = = 11 + + Δ Δ {L L}_{33}

Calculate the arc length discriminant ΔL ₃ of the eigenvalue vector of the centroid distribution of each spot; where: L′ ₃ and L ₃ are the point noise in the image collected by the CCD camera at the previous moment and the next moment of the current moment respectively the arc length parameter of the eigenvalue vector of the distribution;

Step 4. According to the spot centroid distribution image obtained in step 1, for each spot, use the formula:

\frac{\sqrt{{(({x x}_{i i 11} - - {x x}_{i i 22}))}^{22} + + {(({y the y}_{i i 11} - - {y the y}_{i i 22}))}^{22}}}{n no \cdot &Center Dot; \frac{11}{{f f}_{CCD CCD}}} = = 11 + + ΔC ΔC

Calculate the discriminant ΔC of the centroid coordinate change rate of the spot; where (x _i1 , y _i1 ) is the centroid coordinate of the i-th spot at the initial moment, (x _i2 , y _i2 ) is the spot obtained in the n-th sampling period Centroid coordinates, K is the number of spots in the field of view of the CCD camera, i and n are both positive integers;

Step 5. Traversing each spot in the spot centroid distribution image to identify the beacon light. The identification process for each spot is: judging the semicircle area discriminant ΔS ₃ of the eigenvalue vector of the centroid distribution of the spot, and calculating the spot centroid Whether the arc length discrimination quantity ΔL ₃ of the distributed eigenvalue vector is less than 5 pixels, if the judgment result is yes, then execute step 51; if the judgment result is no, then execute step 52;

Step 51. Determine whether the discriminant value ΔC of the centroid coordinate change rate of the spot is less than five pixels per second. If the judgment result is yes, the spot is a star background spot. When all the spots traversed are stellar background spots , indicating that there is no beacon light spot in the image at the current moment, go to step 6; if the judgment result is no, go to step 52;

Step 52: Compare the discriminant amount ΔC of the centroid coordinate change rate of all light spots, and the spot corresponding to the largest discriminant amount ΔC of the centroid coordinate change rate is the beacon light spot at the current moment, so as to realize beacon light recognition and noise reduction, and Execute step six;

Step 6. Use the image corresponding to the next moment as the image at the current moment, and return to step 1 to continue beacon light recognition.