CN112986947B - Machine learning-based trace point filtering processing method - Google Patents
Machine learning-based trace point filtering processing method Download PDFInfo
- Publication number
- CN112986947B CN112986947B CN202110392273.5A CN202110392273A CN112986947B CN 112986947 B CN112986947 B CN 112986947B CN 202110392273 A CN202110392273 A CN 202110392273A CN 112986947 B CN112986947 B CN 112986947B
- Authority
- CN
- China
- Prior art keywords
- trace
- clutter
- radar
- point
- trace point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 24
- 238000010801 machine learning Methods 0.000 title claims abstract description 13
- 238000003672 processing method Methods 0.000 title claims abstract description 11
- 238000013179 statistical model Methods 0.000 claims description 19
- 238000001514 detection method Methods 0.000 claims description 12
- 238000000034 method Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000009825 accumulation Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000013139 quantization Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/414—Discriminating targets with respect to background clutter
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a trace point filtering processing method based on machine learning, which is used for establishing a nonparametric model for real-time trace point data and calculating the optimal trace point filtering parameter characteristics matched with the actual trace point data. The invention adopts real-time trace big data to accumulate trace point number and trace point amplitude modeling estimation in different areas, realizes trace point filtration by utilizing the estimation value, has high precision and high reliability, does not need manual intervention, and improves the intelligent level of the system.
Description
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to a trace point filtering processing method based on machine learning.
Background
After the radar detection signal is processed, more strong clutter echo point traces still remain, and the difficulty of target detection and tracking is increased. In the data processing process, residual traces need to be subjected to trace filtering processing according to target echo trace characteristics, such as echo width, distance width and other traditional characteristics, so as to eliminate clutter traces. The target echo point trace characteristics are influenced by various factors such as a signal detection algorithm, clutter characteristics and the like, the echo point trace statistical characteristics are unstable, the robustness is poor, and manual intervention and optimization are needed.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to solve the defects in the prior art and provides a machine learning-based trace point filtering processing method, which effectively identifies a target trace point and a clutter point by using the statistical characteristics of the average value and the variance of the trace point amplitude calculated in real time by using multi-cycle trace points.
The technical scheme is as follows: the invention discloses a trace point filtering processing method based on machine learning, which comprises the following steps:
step S1: the radar carries out multi-cycle scanning detection, a Cartesian coordinate system is established for the clutter map, and then corresponding configuration parameters are set;
EchoMapCfg.xMinthe minimum distance in the X direction of the radar is obtained;EchoMapCfg.xMaxthe maximum distance in the X direction of the radar is obtained;EchoMapCfg.yMinis the minimum distance in the Y direction of the radar;EchoMapCfg.yMinis the minimum distance in the Y direction of the radar;EchoMapCfg.xUnitthe distance quantized value in the X direction of the radar is obtained;EchoMapCfg.yUnitdistance quantization values in the Y direction of the radar are obtained;EchoMapCfg.DetectPeriodupdating a period for the clutter map;EchoMapCfg.Enableeffectively marking clutter maps;
step S2: carrying out gridding processing on a radar detection range according to a radar X-direction distance quantized value and a radar Y-direction distance quantized value in a Cartesian coordinate system;
step S3: mapping the point traces sequentially detected in each scanning period of the radar into corresponding clutter map grids, accumulating the number of the point traces in the same grid on the corresponding clutter map, and storing the amplitude of the point traces;
step S4: if the clutter map is effective, filtering the point traces in the strong clutter grid area, otherwise executing the step S5;
if the amplitude of the echo point trace meets the following formula, the point trace is a clutter point trace, the point trace is filtered, and otherwise, the point trace information is kept.
Step S5: constructing a nonparametric statistical model to calculate the number of traces, if the radar scanning period counting value meets the clutter map updating condition, updating the region information of the clutter map, and setting the clutter map to be in an effective state; meanwhile, resetting the clutter map updating period to zero;
step S6: for a strong clutter grid region, ordering trace point amplitudes in grids of the strong clutter grid region from small to large, establishing a non-parameter statistical model, and calculating a filtering threshold value of the trace point amplitudes;
step S7: and repeating the steps S3-S6 until the radar does not operate any more. That is, the clutter map is updated in real time as long as the radar is running, and the clutter region is continuously identified by updating (for example, the motion of the cloud and rain weather can change the clutter region).
To ensure that each grid size is consistent, the formula of the grid in step S2 is:
xMaxUnitandyMaxUnitthe maximum number of grids in the X-direction distance and the Y-direction distance, respectively.
Further, the detailed method of step S3 is:
step S3.1, setting a radar scanning period counting valueScanCntOf 1 atiThe echo trace information is expressed as (PlotDis, PlotAzi,PlotAmp)；
Wherein,PlotDisin order to be the distance of the point trace,PlotAziin order to determine the position of the point trace,PlotAmpis the trace point amplitude;
calculating the echo traceiMesh numbering in a clutter mesh mapIndexThe calculation formula is as follows,
s3.2, performing trace point accumulation updating on the clutter grids corresponding to the trace points, and simultaneously storing trace point amplitude information: the method of accumulation update is as follows:
Cntthe count value of the trace point amplitude in the clutter grid is obtained, and the initialization value is zero.
Further, the detailed method of step S5 is:
s5.1, if the radar scanning period is satisfiedScanCnt = EchoMapCfg.DetectPeriodUpdating the clutter map according to the following steps;EchoMapCfg.DetectPeriodis the period of clutter map update; updating the clutter map once in 15 cycles, for example, and setting according to experience;
s5.2, accumulating point trace data structure for clutter map gridsstPlotDensityThe trace point numbers (non-zero values) in the array are sorted from small to large, and the effective sample number after sorting isPlotSize；
S5.3, pairstPlotDensityThe middle accumulated trace point number is used for establishing a non-parametric statistical model, and the window width of the non-parametric statistical model isWndThe number of effective samples isPlotSize(ii) a Then assume each accumulated trace countmCalculating by Gaussian statistical model as mean valuestPlotDensityAll accumulated trace pointsm’For accumulated trace point numbermProbability of (2)
(ii) a Wherein the mean value is the accumulated trace numbermThe resolution width is the width of the model window;
mandm’ =1,2,..., PlotSize；
s5.4, obtaining an arrayprobThe accumulated trace point number corresponding to the maximum probability value in (1) is the average value of the accumulated trace point numbersPlotAvrAnd calculating the variance of accumulated trace pointsPlotStd；
S5.5, according to the average value of the accumulated trace pointsPlotAvrAnd accumulated trace point number variancePlotStdSetting a strong clutter region trace point threshold of a clutter mapEchoPlotDensityNon-clutter region trace point thresholdNonEchoPlotDensity(ii) a The calculation is shown in the following formula,
wherein,Coffis a proportionality coefficient, set according to experience;
s5.6, updating the clutter map grid region, and identifying a strong clutter region, a weak clutter region and a non-clutter region based on accumulated trace point number thresholds of different grid regions, wherein the judgment conditions are as follows:
wherein,tthe clutter map grid regions are numbered,t=1,2,...,xMaxUnit*yMaxUnit。
further, the detailed method of step S6 is:
s6.1, traversing clutter map grid regiontIf a gridtIf the area mark is a strong clutter area, executing the step S6.2, otherwise, continuously traversing;t=1,2,...,xMaxUnit*yMaxUnit；
s6.2, carrying out point trace amplitude array on grid region of strong clutter mapstPlotAmp().AmpAccording to the smooth arrangement of the amplitudes from small to large, the effective amplitude samples of the point traces in the clutter grid after sequencing are counted asCnt；
S6.3, establishing a non-parametric statistical model for the trace amplitude, wherein the model window width of the non-parametric statistical model isWndThe sample is counted asCnt(ii) a Successively assuming each trace amplitudeuCalculating all amplitudes by using a Gaussian statistical model as a mean valueu'Trace width of pointuProbability of (2)prob(u)Wherein the mean is the trace point amplitudeuThe resolution width is the window width;
uandu’=1,2,,,,,Cnt；
s6.4, obtaining an arrayprobThe maximum value of the probability in (1) corresponds to the trace point amplitude, and the value is the mean value of the trace point amplitudesAmpAvrAnd calculating the variance of trace point amplitudeAmpStd；
S6.5, according to the mean value of the trace point amplitudesAmpAvrAnd trace amplitude varianceAmpStdCalculating the trace point amplitude threshold of the strong clutter regionAmpLit；
The calculation is shown in the following formula,
wherein,Coff2is a scaling factor, set empirically.
Has the advantages that: establishing a nonparametric model for real-time trace point data, and calculating optimal trace point filtering parameter characteristics matched with the actual trace point data through the nonparametric model; compared with the prior art, the method has the advantages of high precision, high reliability, certain robustness, no need of manual intervention and extremely high intelligence level.
Drawings
FIG. 1 is a schematic overall flow diagram of the present invention;
FIG. 2 is a schematic diagram illustrating region identification in an embodiment;
FIG. 3 is a diagram illustrating an embodiment of a filter capable of trace-dropping;
FIG. 4 is a diagram illustrating trace point count changes before and after trace point filtering processing in the embodiment.
Detailed Description
The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.
As shown in fig. 1, the invention utilizes radar multi-cycle accumulation point traces to establish a clutter map, and automatically identifies the area characteristics of a strong clutter area, a weak clutter area and a non-clutter area detected by the radar; and then filtering the traces in different areas, namely: and establishing a nonparametric statistical model of the trace point amplitude based on machine learning, and then calculating the mean value and the variance of the trace point amplitude in the corresponding region through the nonparametric statistical model.
The invention is based on machine learning to count the accumulated point trace number threshold of the strong clutter region, the weak clutter region and the non-clutter region in real time, and the clutter amplitude filtering threshold of the strong clutter region, without manually setting the threshold.
The method comprises the following specific steps:
step 1: the radar carries out multi-cycle scanning detection, a Cartesian coordinate system is established for the clutter map, and then corresponding configuration parameters are set; as shown in fig. 2, the periodic radar scan result in the present embodiment is: the green area is echo point traces in a non-clutter area, the blue area is weak clutter area point traces, and the black area is clutter area point traces;
step 2: in a Cartesian coordinate system, gridding a radar detection range according to a radar X-direction distance quantized value and a radar Y-direction distance quantized value;
and step 3: mapping the point traces sequentially detected in each scanning period of the radar into corresponding clutter map grids, accumulating the number of the point traces in the grids, and storing the amplitude of the point traces;
and 4, step 4: if the clutter map is effective, filtering the point traces in the strong clutter grid area, otherwise executing the step 5; as shown in fig. 3, the green area is the actual detection trace of the radar, and the red is the unfiltered echo trace;
and 5: if the radar scanning period counting value meets the clutter map updating condition, updating the clutter map region information, and setting the clutter map to be in an effective state; meanwhile, the updating period of the clutter map is reset to zero;
step 6: for a strong clutter grid region, sequencing trace point amplitudes in a grid from small to large, establishing a non-parameter statistical model, and calculating a filtering threshold value of the trace point amplitudes;
and 7: and repeating the step 3 to the step 6.
In the embodiment, the radar detection point trace number and the unfiltered echo point trace number are compared in multiple cycles, as shown in fig. 4, the unfiltered echo point trace is obviously reduced after being filtered by the technical scheme of the invention.
Because the target amplitude characteristic and the background clutter amplitude characteristic in the clutter region are different, the embodiment can show that the point trace filtering in different regions can be realized, a large amount of residual point traces after signal processing can be reduced, and the target detection and tracking capability of subsequent data processing can be improved.
Claims (5)
1. A trace point filtering processing method based on machine learning is characterized in that: the method comprises the following steps:
step S1: the radar carries out multi-cycle scanning detection, a Cartesian coordinate system is established for the clutter map, and then corresponding configuration parameters are set and initialized;
EchoMapCfg.xMinthe minimum distance in the X direction of the radar is obtained;EchoMapCfg.xMaxthe maximum distance in the X direction of the radar is obtained;EchoMapCfg.yMinis the minimum distance in the Y direction of the radar;EchoMapCfg.yMaxthe maximum distance in the Y direction of the radar;EchoMapCfg.xUnitthe distance quantized value in the X direction of the radar is obtained;EchoMapCfg.yUnitdistance quantization values in the Y direction of the radar are obtained;EchoMapCfg.DetectPeriodupdating a period for the clutter map;
EchoMapCfg.Enableeffectively marking clutter maps;
step S2: carrying out gridding processing on a radar detection range according to a radar X-direction distance quantized value and a radar Y-direction distance quantized value in a Cartesian coordinate system;
step S3: mapping the point traces sequentially detected in each scanning period of the radar into corresponding clutter map grids, accumulating the number of the point traces in the same grid on the corresponding clutter map, and storing the amplitude of the point traces;
step S4: amplitude of trace points in the if clutter grid
Greater than or equal to the trace point amplitude threshold
Then filtering the trace point; otherwise, executing step S5;
step S5: constructing a nonparametric statistical model to calculate the number of traces, if the radar scanning period counting value meets the clutter map updating condition, updating the region information of the clutter map, and setting the clutter map to be in an effective state; meanwhile, resetting the clutter map updating period to zero;
step S6: for the strong clutter grid region, ordering the trace point amplitudes in the grid from small to large, establishing a non-parameter statistical model, and calculating the trace point amplitude threshold value
；
Step S7: and repeating the steps S3-S6 until the radar does not operate any more.
2. The machine learning-based trace point filtering processing method according to claim 1, characterized in that: the formula of the grid in step S2 is:
xMaxUnitandyMaxUnitthe maximum number of grids in the radar X-direction distance and the radar Y-direction distance, respectively.
3. The machine learning-based trace point filtering processing method according to claim 1, characterized in that: the detailed method of the step S3 is as follows:
step S3.1, setting a radar scanning period count value ScanCntiThe echo trace information is expressed as (PlotDis, PlotAzi,PlotAmp)；
Wherein,PlotDisin order to be the distance of the point trace,PlotAziin order to determine the position of the point trace,PlotAmpis the trace point amplitude;
calculating the echo traceiMesh numbering in a clutter mesh mapIndexThe calculation formula is as follows,
s3.2, performing trace point accumulation updating on the clutter grids corresponding to the trace points, and simultaneously storing trace point amplitude information: the method of accumulation update is as follows:
Cntthe count value of the trace point amplitude in the clutter grid is obtained, and the initialization value is zero;
xMaxUnitandyMaxUnitthe maximum number of grids in the radar X-direction distance and the radar Y-direction distance, respectively.
4. The machine learning-based trace point filtering processing method according to claim 1, characterized in that: the detailed method of the step S5 is as follows:
s5.1, if the radar scanning period is satisfiedScanCnt = EchoMapCfg.DetectPeriodUpdating the clutter map according to the following steps;EchoMapCfg.DetectPeriodis the period of clutter map update;
s5.2, accumulating point trace data structure for clutter map gridsstPlotDensityThe trace points in the array are sorted from small to large, and the effective sample number after sorting isPlotSize；
S5.3, pairstPlotDensityThe middle accumulated trace point number is used for establishing a non-parametric statistical model, and the window width of the non-parametric statistical model isWndThe number of effective samples isPlotSize(ii) a Then assume each accumulated trace countmCalculating by Gaussian statistical model as mean valuestPlotDensityAll accumulated trace pointsm’For accumulated trace point numbermProbability of (2)
(ii) a Wherein the mean value is the accumulated trace numbermThe resolution width is the width of the model window;
mandm’ =1,2,..., PlotSize；
s5.4, obtaining an arrayprobThe accumulated trace point number corresponding to the maximum probability value in (1) is the average value of the accumulated trace point numbersPlotAvrAnd calculating the variance of accumulated trace pointsPlotStd；
S5.5, according to the average value of the accumulated trace pointsPlotAvrAnd accumulated trace point number variancePlotStdSetting a strong clutter region trace point threshold of a clutter mapEchoPlotDensityNon-clutter region trace point thresholdNonEchoPlotDensity(ii) a The calculation is shown in the following formula,
wherein,Coffis a scaling factor;
s5.6, updating the clutter map grid region, and identifying a strong clutter region, a weak clutter region and a non-clutter region based on accumulated trace point number thresholds of different grid regions, wherein the judgment conditions are as follows:
whereintThe clutter map grid regions are numbered,t=1,2,...,xMaxUnit*yMaxUnit；xMaxUnitandyMaxUnitthe maximum number of grids in the radar X-direction distance and the radar Y-direction distance, respectively.
5. The machine learning-based trace point filtering processing method according to claim 1, characterized in that: the detailed method of the step S6 is as follows:
s6.1, traversing clutter map grid regiontIf a gridtIf the area mark is a strong clutter area, executing the step S6.2, otherwise, continuously traversing;t=1,2,...,xMaxUnit*yMaxUnit；
s6.2, carrying out point trace amplitude array on grid region of strong clutter mapstPlotAmp(t).AmpAccording to the smooth arrangement of the amplitudes from small to large, the effective amplitude samples of the point traces in the clutter grid after sequencing are counted asCnt；
S6.3, establishing a non-parametric statistical model for the trace amplitude, wherein the model window width of the non-parametric statistical model isWndThe sample is counted asCnt(ii) a Successively assuming each trace amplitudeuCalculating all amplitudes by using a Gaussian statistical model as a mean valueu'Trace width of pointuProbability of (2)prob(u)Wherein the mean is the trace point amplitudeuThe resolution width is the window width;
uandu’=1,2,…,Cnt；
s6.4, obtaining an arrayprobThe maximum value of the probability in (1) corresponds to the trace point amplitude, and the value is the mean value of the trace point amplitudesAmpAvrAnd calculating the variance of trace point amplitudeAmpStd；
S6.5, according to the mean value of the trace point amplitudesAmpAvrAnd trace amplitude varianceAmpStdCalculating the trace point amplitude threshold of the strong clutter regionAmpLit；
The calculation is shown in the following formula,
wherein,Coff2is a scaling factor;xMaxUnitandyMaxUnitthe maximum number of grids in the radar X-direction distance and the radar Y-direction distance, respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110392273.5A CN112986947B (en) | 2021-04-13 | 2021-04-13 | Machine learning-based trace point filtering processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110392273.5A CN112986947B (en) | 2021-04-13 | 2021-04-13 | Machine learning-based trace point filtering processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112986947A CN112986947A (en) | 2021-06-18 |
| CN112986947B true CN112986947B (en) | 2021-07-23 |
Family
ID=76337998
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110392273.5A Active CN112986947B (en) | 2021-04-13 | 2021-04-13 | Machine learning-based trace point filtering processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112986947B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113721211B (en) * | 2021-06-26 | 2024-08-02 | 中国船舶重工集团公司第七二三研究所 | Sparse fixed clutter recognition method based on point trace characteristic information |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104237861A (en) * | 2014-09-15 | 2014-12-24 | 电子科技大学 | Method for obtaining CFAR detection threshold in unknown clutter background |
| CN104793200A (en) * | 2015-04-09 | 2015-07-22 | 电子科技大学 | Dynamic planning track-before-detect method based on iterative processing |
| CN106646396A (en) * | 2016-11-18 | 2017-05-10 | 西安电子工程研究所 | Radar clutter map detection threshold adaptive setting method |
| CN107300698A (en) * | 2017-08-21 | 2017-10-27 | 哈尔滨工业大学 | A kind of Radar Target Track initial mode based on SVMs |
| CN110161464A (en) * | 2019-06-14 | 2019-08-23 | 成都纳雷科技有限公司 | A kind of Radar Multi Target clustering method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101551811B1 (en) * | 2014-05-19 | 2015-09-10 | 최수호 | Radar apparatus and method for frequency interference cancellation thereof |
-
2021
- 2021-04-13 CN CN202110392273.5A patent/CN112986947B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104237861A (en) * | 2014-09-15 | 2014-12-24 | 电子科技大学 | Method for obtaining CFAR detection threshold in unknown clutter background |
| CN104793200A (en) * | 2015-04-09 | 2015-07-22 | 电子科技大学 | Dynamic planning track-before-detect method based on iterative processing |
| CN106646396A (en) * | 2016-11-18 | 2017-05-10 | 西安电子工程研究所 | Radar clutter map detection threshold adaptive setting method |
| CN107300698A (en) * | 2017-08-21 | 2017-10-27 | 哈尔滨工业大学 | A kind of Radar Target Track initial mode based on SVMs |
| CN110161464A (en) * | 2019-06-14 | 2019-08-23 | 成都纳雷科技有限公司 | A kind of Radar Multi Target clustering method and device |
Non-Patent Citations (1)
| Title |
|---|
| 基于杂波特征评估的雷达目标点迹过滤方法;史建涛等;《太赫兹科学与电子信息学报》;20191231;第17卷(第6期);第988-933页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112986947A (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110991272B (en) | Multi-target vehicle track recognition method based on video tracking | |
| CN109272489B (en) | Infrared weak and small target detection method based on background suppression and multi-scale local entropy | |
| CN109325935B (en) | Power transmission line detection method based on unmanned aerial vehicle image | |
| CN103632382B (en) | A kind of real-time multiscale target tracking based on compressed sensing | |
| CN107292310B (en) | Visual positioning and automatic reading method for circular pointer type dial plate | |
| CN111862194B (en) | Deep learning plant growth model analysis method and system based on computer vision | |
| CN104112269B (en) | A kind of solar battery laser groove parameter detection method and system based on machine vision | |
| CN111563469A (en) | Method and device for identifying irregular parking behaviors | |
| CN111007531A (en) | Road edge detection method based on laser point cloud data | |
| CN111598780B (en) | Terrain adaptive interpolation filtering method suitable for airborne LiDAR point cloud | |
| CN104063711B (en) | A kind of corridor end point fast algorithm of detecting based on K means methods | |
| CN109708658B (en) | Visual odometer method based on convolutional neural network | |
| CN112200160A (en) | Deep learning-based direct-reading water meter reading identification method | |
| CN107301649B (en) | Regional merged SAR image coastline detection algorithm based on superpixels | |
| CN112986947B (en) | Machine learning-based trace point filtering processing method | |
| CN111242000A (en) | Road edge detection method combining laser point cloud steering | |
| CN109359604A (en) | Meter recognition method under shadow interference towards crusing robot | |
| CN113326925A (en) | Density graph regression-based flower amount statistical method, system, equipment and medium for convolutional neural network | |
| CN114881960A (en) | Feature enhancement-based cloth linear defect detection method and system | |
| CN106815851B (en) | A kind of grid circle oil level indicator automatic reading method of view-based access control model measurement | |
| CN114494695A (en) | Intelligent water conservancy urban and rural waterlogging level monitoring and early warning method and device | |
| CN110633679B (en) | Automatic pointer instrument indicating identification method and system based on genetic algorithm | |
| CN111880197A (en) | Road boundary detection method based on four-line laser radar | |
| CN116977893A (en) | Shoal state detection method based on improved sobel and piecewise linear transformation | |
| CN111339824A (en) | Road surface sprinkled object detection method based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 210000 Room 301, floor 3, building 75, zone B, entrepreneurship and innovation city, No. 15, Fengji Avenue, Yuhuatai District, Nanjing, Jiangsu Province Patentee after: Nanjing Thunderbolt Information Technology Co.,Ltd. Address before: 210000 building 75, area B, Chuangye new town, No.15 Fengji Avenue, Yuhuatai District, Nanjing City, Jiangsu Province Patentee before: NANJING LEADING INFORMATION TECHNOLOGY Co.,Ltd. |