Pplot quality evaluation method based on analytic hierarchy process

Abstract

The invention belongs to the field of radar signal processing, and particularly relates to a trace point quality evaluation method based on an analytic hierarchy process. According to a large number of samples of a correct target trace point block, statistical characteristics of probability density functions of various characteristics (distance broadening, azimuth broadening, Doppler channel stability, a signal-to-noise ratio and a detection point number ratio) of the trace point block are analyzed; according to the probability density function of each feature, calculating to obtain the confidence coefficient of the corresponding feature of each trace point block; and calculating a weight value corresponding to each feature in combination with a chromatographic analysis method, carrying out weighted summation on each feature, and comprehensively obtaining quality evaluation of the trace point block. According to the method, target trace point extraction and clutter trace point suppression are realized, and meanwhile, the method can be used for target and non-target judgment and association in the track association process.

Description

Technical Field

[0001] The invention belongs to the field of radar signal processing, and specifically relates to a point trace quality assessment method based on a hierarchical analysis method, which is suitable for target point trace extraction and clutter point trace suppression in a point trace processing process, and can also be used for judging and associating targets and non-targets in a track association process. Background Art

[0002] After the radar completes signal processing for the received echo signal, it obtains many detection points, among which there are a large number of clutter detection points of different attribute categories and other multi-target detection points. After the point trace condensation process, the centroid of the point trace is obtained for the back-end data processing process. Due to the existence of clutter point traces, a large number of clutter points and possible target points appear in the wave gate, which brings a computational burden to the data association process and generates false tracks. Therefore, the point trace quality evaluation is performed in the point trace processing process, and based on the evaluation results, some non-target point traces are eliminated, and the quality information is transmitted to the terminal data processing module to provide a basis for data association. Summary of the invention

[0003] For the evaluation of the quality of the points after condensation, the present invention proposes a point quality evaluation method based on the hierarchical analysis method. According to a large number of samples of the correct target point blocks, the statistical characteristics of the probability density function of various features of this type of point block (range broadening, azimuth broadening, Doppler channel stability, signal-to-noise ratio and detection point number ratio) are analyzed; according to the probability density function of each feature, the confidence of the corresponding feature of each point block is calculated; combined with the tomographic analysis method, the corresponding weight of each feature is calculated, and the weighted sum of each feature is performed to comprehensively obtain the quality evaluation of the point block.

[0004] The technical solution of the present invention is: a point trace quality evaluation method based on hierarchical analysis method.

[0005] Step 1: Collect a large number of correct target point trace samples (non-clutter samples) and fit the probability density function of various features of this type of point trace block.

[0006] The probability density functions of four types of features, namely, range broadening, azimuth broadening, Doppler channel stability in all azimuths of the range unit where the maximum amplitude detection point is located, and the ratio of the maximum signal-to-noise ratio to the number of detection points, of all point trace blocks of the collected correct targets are statistically analyzed and the data are saved.

[0007] Step 2: Calculate the confidence of the four types of features of each point block.

[0008]

[0009] In the formula, θ i is the confidence of the i-th feature; is the probability density value corresponding to the ith eigenvalue of the point trace block obtained according to the probability density function of the ith feature; is the maximum probability density value of the i-th feature probability density function, where i = 1, 2, 3, 4.

[0010] Step 3: Based on the hierarchical analysis method, calculate the weight corresponding to each feature.

[0011] Use numbers to quantify the importance of features:

[0012]

[0013] Construct a 4-order matrix based on the four types of features of the point trace block:

[0014]

[0015] Where: a ij The importance scale of factor i relative to factor j is represented by the values ​​1 to 9 in the above table.

[0016] According to the matrix A, the weight values ​​W=(w1,w2,w3,w4)′ corresponding to the four types of features are calculated as follows.

[0017] (1) First, the obtained matrix A is normalized by column, that is, each element in A is divided by the sum of its column, and the matrix is ​​obtained

[0018]

[0019] snm1=a 11 +1 / a 12 +1 / a 13 +1 / a 14

[0020] snm2=a 12 +a 22 +1 / a 23 +1 / a 13

[0021] snm3=a 13 +a 23 +a 33 +1 / a 34

[0022] snm4=a 14 +a 24 +a 34 +a 44

[0023] in, It represents the result of normalizing the A matrix by column; sum1 represents the sum of all elements in the first column of the A matrix. Similarly, snm2, snm3, and snm4 represent the sum of all elements in the corresponding columns of the A matrix.

[0024] (2) Secondly, the obtained Add the matrices row by row and divide the result of each addition by The number of columns of the matrix is ​​the weight value W:

[0025]

[0026] Among them, w1 represents the weight of the point trace distance broadening feature, w2 represents the weight of the point trace azimuth broadening feature, w3 represents the weight of the point trace Doppler channel stability feature, and w4 represents the weight of the point trace signal-to-noise ratio and the number of detection points ratio feature; express The sum of all column elements in the first row of the matrix. express The sum of all column elements in the corresponding row of the matrix; num represents The number of matrix columns.

[0027] Step 4: Calculate the quality of each point block.

[0028] Combine the confidence of the four types of features of a certain point obtained in step 2 with the feature weight W obtained in step 3 to calculate the comprehensive quality of the point block.

[0029]

[0030] In the formula, θ i is the confidence of the i-th feature of the point trace block; w i The weight corresponding to the i-th feature, where i = 1, 2, 3, 4; quality is the comprehensive quality of the current point trace block.

[0031] The present invention starts from the perspective of target point trace features and combines the hierarchical analysis method to achieve objective quality evaluation of each point trace after condensation. First, due to the large number of types of clutter, including ground clutter, sea clutter, meteorological clutter and echoes of other targets in the surrounding environment, and these clutters vary greatly with different environments, it is difficult to perform feature statistics on clutter point trace blocks, and the distribution range of the grid feature probability density function is large and not representative. Considering that the characteristic distribution characteristics of the target point trace block are relatively stable and concentrated, the present invention starts from the perspective of target characteristics and compares and comprehensively judges the various types of point traces after condensation. Secondly, considering that some clutter point traces and target point traces have a large similarity in a certain feature, the present invention adopts a method of comprehensive judgment of multiple features, which draws on the idea of ​​multi-feature joint probability density, and can more accurately distinguish the target and clutter point traces. Finally, using the point trace quality evaluation results, the clutter point traces can be judged and suppressed, which reduces the workload of the data processing link to a certain extent and improves the accuracy of data association. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is the probability distribution diagram of the distance broadening of the correct target point trace samples.

[0033] Figure 2 It is the probability distribution diagram of the correct target point sample position broadening.

[0034] Figure 3 It is the Doppler channel consistency probability distribution diagram of the correct target point trace samples.

[0035] Figure 4 It is the probability distribution diagram of the correct target point trace samples (signal-to-noise ratio / detection point).

[0036] Figure 5 Displays the radar echo detection point.

[0037] Figure 6 Display of detected points in polar coordinates before suppressing point traces

[0038] Figure 7 Display of detection points in polar coordinates after point trace suppression DETAILED DESCRIPTION

[0039] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0040] The present invention provides a point trace quality assessment method based on hierarchical analysis method, which can objectively assess the quality of a point trace and has a certain filtering ability for non-target points traces.

[0041] The experimental process of this embodiment is based on the measured data of the near-coast detection of the dual-band radar. The point trace condensation and point trace quality evaluation are completed for the detection video information obtained after signal processing. Assume that the total number of point trace samples obtained by point trace condensation is R, of which the correct aircraft target sample is B and the other clutter samples are C, R = B∪C.

[0042] Step 1: Collect a large number of correct target point trace samples B, and fit the probability density function of the four characteristics of this type of point trace block (range broadening, azimuth broadening, Doppler channel stability, signal-to-noise ratio and ratio of the number of detection points).

[0043] pdf dis_around =P(B dis_around )

[0044] pdf azi_around =P(B azi_around )

[0045] pdf doupler =P(B doupler )

[0046] pdf snr / Ep =P(B snr / Ep )

[0047] Among them, pdf dis_around represents the probability density distribution of the point track distance broadening; pdf azi_around represents the probability density distribution of the point trace azimuth broadening; pdf doupler Represents the probability density distribution of the Doppler channel stationarity of the point trace; pdf snr / Ep Represents the probability density distribution of the signal-to-noise ratio of the point trace and the ratio of the number of detection points; B dis_around represents the distance stretching sample composed of the distance stretching value of each point trace in the target point trace sample B; B azi_around represents the azimuth stretch sample composed of the azimuth stretch value of each point trace in the target point trace sample B; B doupler represents the Doppler channel stability sample composed of the Doppler channel stability value of each point in the target point sample B; B snr / Ep represents the sample of the set signal-to-noise ratio and the number of detection points of each point in the target point sample B; P(·) represents the discrete feature sample set (B dis_around , B azi_around , B doupler , B snr / Ep ) is mapped to the probability density distribution function of the corresponding features.

[0048] Step 2: Calculate the confidence of the four types of features of each point in the total point sample R.

[0049] Assume that the distance spread of a point k is k dis , the azimuth spread is k azi , the Doppler channel stationarity size is k douple , the ratio of signal-to-noise ratio to the number of detection points is k snr / Ep . Find the probability density value corresponding to a certain feature from the probability density distribution corresponding to the feature value.

[0050]

[0051] Calculate the maximum probability density value corresponding to the probability density distribution of four types of characteristic attributes (range broadening, azimuth broadening, Doppler channel stability, signal-to-noise ratio and ratio of detection points) respectively

[0052] Where i represents the i-th characteristic attribute (range spread, azimuth spread, Doppler channel stability, signal-to-noise ratio and ratio of detection points), i = 1, 2, 3, 4; k i Indicates the numerical value of the i-th characteristic attribute of a point k, The horizontal axis value representing the probability density distribution of the i-th type of feature attribute; The probability density value corresponding to the value of the i-th characteristic attribute of a point k.

[0053] According to the probability density values ​​of the four types of features of a certain point k and the maximum probability density value in the probability density function of the corresponding features, the confidence of the four types of features of the point k is calculated.

[0054]

[0055] In the formula, θ i is the confidence of the i-th feature of point k; The probability density value corresponding to the value of the i-th characteristic attribute of a certain point k; is the maximum probability density value of the i-th feature probability density function, where i = 1, 2, 3, 4.

[0056] Step 3: Based on the hierarchical analysis method, calculate the weights corresponding to each of the four types of features.

[0057] Use numbers to quantify the importance of features:

[0058]

[0059] Construct a 4-order matrix based on the four types of features of the point trace block:

[0060]

[0061] Where: a ijThe importance scale of factor i relative to factor j is represented by the values ​​1 to 9 in the above table.

[0062] According to the matrix A, the weight values ​​W=(w1,w2,w3,w4)′ corresponding to the four types of features are calculated as follows.

[0063] (1) First, the obtained matrix A is normalized by column, that is, each element in A is divided by the sum of its column, and the matrix is ​​obtained

[0064]

[0065] snm1=a 11 +1 / a 12 +1 / a 13 +1 / a 14

[0066] snm2=a 12 +a 22 +1 / a 23 +1 / a 13

[0067] snm3=a 13 +a 23 +a 33 +1 / a 34

[0068] snm4=a 14 +a 24 +a 34 +a 44

[0069] in, It represents the result of normalizing the A matrix by column; sum1 represents the sum of all elements in the first column of the A matrix. Similarly, snm2, snm3, and snm4 represent the sum of all elements in the corresponding columns of the A matrix.

[0070] (2) Secondly, the obtained Add the matrices row by row and divide the result of each addition by The number of columns of the matrix is ​​the weight value W:

[0071]

[0072] Among them, w1 represents the weight of the point trace distance broadening feature, w2 represents the weight of the point trace azimuth broadening feature, w3 represents the weight of the point trace Doppler channel stability feature, and w4 represents the weight of the point trace signal-to-noise ratio and the number of detection points ratio feature; express The sum of all column elements in the first row of the matrix. express The sum of all column elements in the corresponding row of the matrix; num represents The number of matrix columns.

[0073] Step 4: Calculate the quality of each point block.

[0074] Combine the four types of feature confidence of a certain point k obtained in step 2 with the feature weight W obtained in step 3 to calculate the comprehensive quality of the point block.

[0075]

[0076] In the formula, θ i is the confidence of the i-th feature of the point trace block; w i The weight corresponding to the i-th feature in the four categories of features, where i = 1, 2, 3, 4; quality is the comprehensive quality of the current point k.

[0077] The present invention adopts the hierarchical analysis method to perform comprehensive calculation on the confidence of various physical characteristics of the point trace block, so as to realize the quality evaluation of the point trace. Figure 1 to Figure 4 The probability density distribution of the four types of physical features of the target sample's point trace block is displayed. It can be seen that the feature distribution of the target point trace block is limited to a certain range. By calculating a certain type of feature of the point trace block and comparing it with the probability distribution function of the feature, the target confidence corresponding to the feature can be obtained. Figure 5 This is a display diagram of the detection points of the radar multi-circle detection. It can be seen that multiple batches of targets appear within the detection range, and there is dense clutter interference in the close range, which has a great impact on data association. Figure 6 and Figure 7 , the quality of each condensed point is evaluated, and a quality threshold is set to suppress non-target points. It can be seen that after point suppression, the clutter points are significantly reduced and the target point loss is less.

[0078] The beneficial effects of the present invention are:

[0079] In step 1, for the division of target points and clutter points, from the physical perspective of the point block, various feature probability statistical models are analyzed in detail. Since there are many types of clutter and the point feature probability models of various types of clutter have large distribution differences, it is not possible to establish a unified model for them. Since the various features of the target point trace are related to the signal processing process, if the signal processing process is determined, a unified point trace feature model can be established for the target point trace detected by the radar. In the process of comparing clutter with the target, the established target point trace feature model can be used as an evaluation criterion.

[0080] In step 3, the weights corresponding to the four types of features are calculated using hierarchical analysis. Figure 1 to Figure 4The Doppler channel stationarity probability density of the target point block and the signal-to-noise ratio / detection point number probability density distribution have great centralization, which can determine the range in which most targets are concentrated under this type of feature. In the hierarchical analysis method, the centralization characteristics of the probability density function are used to objectively determine the importance of each type of feature, and then the judgment matrix is ​​obtained, and the weights of each type of feature are calculated.

Claims

1. A point trace quality assessment method based on hierarchical analysis method, characterized by: The following steps are included: Step 1: Collect the correct target point trace sample B and fit the probability density function of the four types of features of this type of point trace block; pdf dis_around JP(B dis_around ) pdf azi_around JP(B azi_around ) pdf doupler JP(B doupler ) pdf snr / Ep JP(B snr / Ep ) Among them, pdf dis_around represents the probability density distribution of the point track distance broadening; pdf azi_around represents the probability density distribution of the point trace azimuth broadening; pdf doupler Represents the probability density distribution of the Doppler channel stationarity of the point trace; pdf snr / Ep Represents the probability density distribution of the signal-to-noise ratio of the point trace and the ratio of the number of detection points; B dis_around represents the distance stretching sample composed of the distance stretching value of each point trace in the target point trace sample B; B azi_around represents the azimuth stretch sample composed of the azimuth stretch value of each point trace in the target point trace sample B; B doupler represents the Doppler channel stability sample composed of the Doppler channel stability value of each point in the target point sample B; B snr / Ep represents the sample of the set signal-to-noise ratio and the ratio of the number of detection points of each point in the target point sample B; P(·) represents the probability density distribution function mapping of the discrete feature sample set and the corresponding feature; Step 2: Calculate the confidence of the four types of features of each point in the total point sample R; Step 3: Based on the hierarchical analysis method, calculate the weights corresponding to the four types of features; Construct a 4-order matrix based on the four types of features of the point trace block: Where: a ij It represents the importance scale of factor i relative to factor j; According to the matrix A, the weight values ​​W=(w1,w2,w3,w4)′ corresponding to the four types of features are calculated; Step 4: Calculate the quality of each point block. In the formula, θ i is the confidence of the i-th feature of the point trace block; w i The weight corresponding to the i-th feature in the four types of features, where i = 1, 2, 3, 4; quality is the comprehensive quality of the current point trace.

2. The method for evaluating the quality of a point trace based on the analytic hierarchy process according to claim 1, characterized in that: The specific calculation process of confidence in step 2 is: Assume that the distance spread of a point k is k dis , the azimuth spread is k azi , the Doppler channel stationarity size is k douple , the ratio of signal-to-noise ratio to the number of detection points is k snr / Ep ; Find the probability density value corresponding to a certain feature from the probability density distribution corresponding to the feature value; Calculate the maximum probability density value corresponding to the probability density distribution of the four types of feature attributes respectively Where i represents the i-th characteristic attribute, i = 1, 2, 3, 4; k i Indicates the numerical value of the i-th characteristic attribute of a point k, pdf x i The horizontal axis value representing the probability density distribution of the i-th type of feature attribute; The probability density value corresponding to the value of the i-th characteristic attribute of a certain point k; According to the probability density values ​​of the four types of features of a certain point k and the maximum probability density value in the probability density function of the corresponding features, the confidence of the four types of features of the point k is calculated; In the formula, θ i is the confidence of the i-th feature of point k; The probability density value corresponding to the value of the i-th characteristic attribute of a certain point k; is the maximum probability density value of the i-th feature probability density function, where i = 1, 2, 3, 4.

3. A point trace quality assessment method based on hierarchical analysis method according to claim 1 or 2, characterized in that: The specific calculation steps of the weight value in step 3 are as follows: (1) Normalize the obtained matrix A by column, that is, divide each element in A by the sum of its column to obtain the matrix snm1=a 11 +1 / a 12 +1 / a 13 +1 / a 14 snm2=a 12 +a 22 +1 / a 23 +1 / a 13 snm3=a 13 +a 23 +a 33 +1 / a 34 snm4=a 14 +a 24 +a 34 +a 44 in, It represents the result of normalizing the A matrix by column; sum1 represents the sum of all elements in the first column of the A matrix, and similarly, snm2, snm3, and snm4 represent the sum of all elements in the corresponding columns of the A matrix; (2) The obtained Add the matrices row by row and divide the result of each addition by The number of columns of the matrix is ​​the weight value W: Among them, w1 represents the weight of the point trace distance broadening feature, w2 represents the weight of the point trace azimuth broadening feature, w3 represents the weight of the point trace Doppler channel stability feature, and w4 represents the weight of the point trace signal-to-noise ratio and the ratio of the number of detection points feature; express The sum of all column elements in the first row of the matrix. express The sum of all column elements in the corresponding row of the matrix; num represents The number of matrix columns.

4. A point trace quality assessment method based on hierarchical analysis method according to claim 1 or 2, characterized in that: The importance scale comparison value in step 3 is: 。

Images