CN113822909B - Water flow velocity measurement method based on motion enhancement features - Google Patents
Water flow velocity measurement method based on motion enhancement features Download PDFInfo
- Publication number
- CN113822909B CN113822909B CN202111157928.7A CN202111157928A CN113822909B CN 113822909 B CN113822909 B CN 113822909B CN 202111157928 A CN202111157928 A CN 202111157928A CN 113822909 B CN113822909 B CN 113822909B
- Authority
- CN
- China
- Prior art keywords
- water flow
- frame difference
- points
- motion
- flow velocity
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 108
- 238000000691 measurement method Methods 0.000 title claims abstract description 20
- 238000000034 method Methods 0.000 claims abstract description 36
- 239000013598 vector Substances 0.000 claims abstract description 17
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 238000013145 classification model Methods 0.000 claims abstract description 14
- 238000006073 displacement reaction Methods 0.000 claims abstract description 10
- 230000000694 effects Effects 0.000 claims abstract description 7
- 230000002708 enhancing effect Effects 0.000 claims abstract description 5
- 238000009432 framing Methods 0.000 claims abstract description 5
- 238000010586 diagram Methods 0.000 claims description 31
- 239000000700 radioactive tracer Substances 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 11
- 208000034699 Vitreous floaters Diseases 0.000 claims description 8
- 238000013135 deep learning Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 230000001629 suppression Effects 0.000 claims description 2
- 238000005286 illumination Methods 0.000 abstract description 9
- 238000012544 monitoring process Methods 0.000 abstract description 8
- 238000005259 measurement Methods 0.000 abstract description 7
- 238000001514 detection method Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 3
- 238000004134 energy conservation Methods 0.000 description 2
- 230000002262 irrigation Effects 0.000 description 2
- 238000003973 irrigation Methods 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 241000271566 Aves Species 0.000 description 1
- 241000238631 Hexapoda Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 210000001503 joint Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01P—MEASURING LINEAR OR ANGULAR SPEED, ACCELERATION, DECELERATION, OR SHOCK; INDICATING PRESENCE, ABSENCE, OR DIRECTION, OF MOVEMENT
- G01P5/00—Measuring speed of fluids, e.g. of air stream; Measuring speed of bodies relative to fluids, e.g. of ship, of aircraft
- G01P5/18—Measuring speed of fluids, e.g. of air stream; Measuring speed of bodies relative to fluids, e.g. of ship, of aircraft by measuring the time taken to traverse a fixed distance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20224—Image subtraction
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Aviation & Aerospace Engineering (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
- Indicating Or Recording The Presence, Absence, Or Direction Of Movement (AREA)
Abstract
The invention discloses a water flow velocity measurement method based on motion enhancement features, which comprises the following steps of collecting and framing water flow video data, calculating a frame difference image by using a frame difference method, counting disturbance noise, filtering the disturbance noise, enhancing the filtered image by using a histogram equalization method to generate water flow feature displacement, dynamically filtering water flow vector outliers, estimating a water flow velocity value, filtering, enhancing and secondarily differentiating continuous multi-frame continuous water flow images to obtain an intermediate result with amplified motion significance, using corner detection and constraint matching for the intermediate result, optimizing on continuous statistical results, obtaining a good real-time measurement effect, and finally utilizing a classification model trained by a collected reference object dataset to further ensure the reliability of a selected reference point of the flow velocity estimation. The method can play a role in more universal water body environment and weaker illumination environment, and is easy to land on the ground of hydrologic intelligent monitoring engineering.
Description
Technical Field
The invention relates to the field of water flow monitoring, in particular to a water flow velocity measurement method based on motion enhancement features.
Background
The water flow speed measurement plays an important role in water resource allocation, agricultural fine irrigation, flood and drought disaster prevention and control and the like. In China with numerous water bodies, numerous rivers, reservoirs and irrigation channels and diversified geological conditions, a more applicable water flow velocity measuring method is found, and the method is very beneficial to actual hydrologic real-time monitoring engineering landing. The non-contact video analysis and measurement method has the advantages of high safety of users, convenience in information butt joint with a background system, difficult damage, capability of collecting more comprehensive information such as image and sound, good real-time performance and the like, and becomes an important ring for intelligent reconstruction of hydrologic monitoring in the future.
At present, a flow rate measuring method based on video analysis has many documents for research and analysis. Based on the pinhole imaging principle, the parameters, the installation position and the machine position angle of the camera can be designed in a self-defined mode, and finally the pixel displacement of the image in unit time can be converted into the physical distance according to a calibration formula. However, a problem is that the current of the water flow in the image is measured by a plurality of pixels, and the problem is solved to a limited extent.
Generally, the existing methods all put a severe requirement on the applicable environment, such as requiring manual tracer throwing, or requiring water flow turbulence with water eddies, or requiring water body itself to have a certain sand content, or requiring a great number of water flow speed calibration sample data sets of different scenes, or requiring more natural floaters to be remarkably presented in the image. In addition, for methods such as template matching, optical flow method, universal version angular point detection and matching, etc., in practice, the requirements of measurement features relied on by the methods on illumination are very high, and in engineering landing, the expected effect is difficult to achieve once images are acquired under the strong light environment.
Many hydrologic monitoring scenes, such as flood discharge monitoring of reservoirs in heavy rain, have not very high requirements on the accuracy of the flow rate, but have high requirements on the time efficiency, and the total amount in a period of time needs to be counted, so that if an image measuring method with strong applicability to various illuminations and water flows and low cost is provided, the method has great significance on practical use.
A simple fact is that the human eye can observe for a period of time in a normal water flow environment with general illumination, locate the movement characteristics which occasionally appear on the water surface, and the characteristics can be water splash bubbles, floaters and the like. Therefore, we propose a water flow velocity measurement method based on motion enhancement features.
Disclosure of Invention
The invention mainly aims to provide a water flow velocity measurement method based on motion enhancement features, which can effectively solve the problems in the background technology.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
a water flow velocity measurement method based on motion enhancement features comprises the following steps:
firstly, acquiring and framing water flow video data, and acquiring water flow surface pictures at the moment t, the moment t+Deltat and the moment t+2Deltat;
step two, using a frame difference method to respectively calculate a frame difference diagram of T1 and a frame difference diagram of T2, wherein T1 is between T time and t+Deltat time, and T2 is between t+Deltat time and t+2 Deltat time;
thirdly, counting disturbance noise, carrying out frame difference processing on the water flow video, counting the fluctuation range of non-zero values of the water flow video, drawing a distribution diagram, taking the highest point as a normal distribution peak, carrying out noise point statistics on the left half side, marking the right side as a maximum noise threshold, and fitting a normal distribution curve;
step four, filtering disturbance noise, denoising the frame difference image according to the self-adaptive threshold calculated in the step three on the frame difference image, filtering random noise which is actually smaller than the self-adaptive threshold, and resetting the random noise;
fifthly, enhancing the filtered picture by using a histogram equalization method, and performing secondary differential processing on the enhanced frame difference picture to obtain a flow velocity characteristic picture with enhanced motion significance characteristics;
step six, generating water flow characteristic displacement, finding out characteristic points by using common local characteristic description operators on a frame difference diagram with enhanced motion significance, and verifying whether a reference object corresponding to the characteristic points is reliable or not by using a classification model;
step seven, dynamically filtering outliers of the water flow vectors, and discarding points out of a range of +/-10% of the highest point in the distribution diagram in the measured value;
and step eight, estimating the flow velocity value of the water flow.
The invention further improves the method, which comprises the following specific steps that firstly, SURF operator is selected to search angular points, the found angular points are traversed by using nearest neighbor matching algorithm to find out preliminary matching points, then partial disordered mismatching points are filtered by using front-back water flow direction consistency and flow speed threshold as constraint conditions, finally, random sampling consistency algorithm RANSAC is used to determine matching characteristic key points, namely, all points which can be matched only want to be very close to one displacement vector, and the vector average value of the characteristic points is the water flow motion vector.
The invention is further improved in that the characteristic points in the step six are judged whether to be reliable tracers or not by a tracer classification model when being selected, and if not, the characteristic points are not selected.
A further improvement of the present invention is that Δt in the step one is the inverse of the camera frame rate.
The invention further improves that the formula for fitting the normal distribution curve in the step three is that
Where d is the frame difference, g (d) represents a gaussian fitted to the frame difference histogram, the standard deviation thereof, and a is the statistical height of the histogram, in practice, the noise with highest occurrence frequency.
The invention further improves that in the fifth step, the flow velocity is inconvenient in a small time interval, the motion significance characteristic has a larger degree of consistency for the two-to-two difference images of three continuous images, and a non-maximum value inhibition processing thought is introduced, so that the most obvious motion characteristic can be highlighted.
The invention further improves that the tracing classification model is based on a general shallow network and trains a cross-water area flow velocity estimation natural tracer classification data set based on deep learning, wherein a data set positive sample comprises natural floaters which are generated by flowing and floaters carried by water bodies; negative examples include non-physical shadow movement effects and self-shifting creatures.
Compared with the prior art, the method can play a role in more universal water environment and weaker illumination environment, is easy for the landing use of hydrologic monitoring intelligent engineering, does not depend on specific water scenes, and can give real-time measurement values every several minutes within 24 hours.
Compared with the prior art, the input of the invention is that a plurality of continuous pictures are obtained from different water flow videos, the output of the pictures is the pixel displacement of key points which can display the flow velocity in the pictures, the natural tracer in the water flow can be automatically marked, and the classification data set can be generated at low cost.
Compared with the prior art, the invention has low illumination requirement, and can measure the river surface by slightly visible naked eyes.
Compared with the prior art, the method has great value in measuring stability and energy conservation and emission reduction of flow velocity measurement at night.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the technical description of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a water flow surface diagram of a water flow velocity measurement method based on motion enhancement features of the present invention.
Fig. 2 is a schematic diagram of a noise distribution curve simulated by a water flow velocity measurement method based on motion enhancement features.
Fig. 3 is a frame difference diagram of a water flow velocity measurement method based on motion enhancement features.
Fig. 4 is a frame difference diagram denoising and filtering effect diagram of a water flow velocity measurement method based on motion enhancement features.
Fig. 5 is a flow velocity profile of an enhanced motion saliency feature of a water flow velocity measurement method based on motion enhancement features of the present invention.
Fig. 6 is a water flow characteristic displacement vector diagram of a water flow velocity measurement method based on motion enhancement characteristics.
Fig. 7 is a diagram showing a measured value distribution diagram of outliers removed by a water flow velocity measurement method based on motion enhancement features according to the present invention.
Fig. 8 is a diagram of the physical flow velocity of water flow based on the method for measuring the flow velocity of water flow with motion enhancement features according to the present invention.
Fig. 9 is an illustration of a natural tracer leaf of a water flow based on a method of measuring velocity of a water flow with motion enhancement features of the present invention.
Fig. 10 is an illustration of a natural tracer spray of a water flow based on a method for measuring velocity of a water flow with motion enhancement features according to the present invention.
Detailed Description
The present invention will be further described with reference to the following detailed description, wherein the drawings are for illustrative purposes only and are shown in schematic drawings, rather than physical drawings, and are not to be construed as limiting the present invention, and in order to better explain the detailed description of the invention, certain components of the drawings may be omitted, enlarged or reduced in size, and not represent the actual product, and it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted, and that all other embodiments obtained by those skilled in the art without making creative efforts fall within the scope of protection of the invention based on the detailed description of the present invention.
Example 1
As shown in fig. 1-7, a water flow velocity measurement method based on motion enhancement features comprises the following steps:
firstly, acquiring and framing water flow video data, and acquiring water flow surface pictures at the moment t, the moment t+Deltat and the moment t+2Deltat;
step two, using a frame difference method to respectively calculate a frame difference diagram of T1 and a frame difference diagram of T2, wherein T1 is between T time and t+Deltat time, and T2 is between t+Deltat time and t+2 Deltat time;
thirdly, counting disturbance noise, carrying out frame difference processing on the water flow video, counting the fluctuation range of non-zero values of the water flow video, drawing a distribution diagram, taking the highest point as a normal distribution peak, carrying out noise point statistics on the left half side, marking the right side as a maximum noise threshold, and fitting a normal distribution curve;
step four, filtering disturbance noise, denoising the frame difference image according to the self-adaptive threshold calculated in the step three on the frame difference image, filtering random noise which is actually smaller than the self-adaptive threshold, and resetting the random noise;
fifthly, enhancing the filtered picture by using a histogram equalization method, and performing secondary differential processing on the enhanced frame difference picture to obtain a flow velocity characteristic picture with enhanced motion significance characteristics;
step six, generating water flow characteristic displacement, finding out characteristic points by using common local characteristic description operators on a frame difference diagram with enhanced motion significance, and verifying whether a reference object corresponding to the characteristic points is reliable or not by using a classification model;
step seven, dynamically filtering outliers of the water flow vectors, and discarding points out of a range of +/-10% of the highest point in the distribution diagram in the measured value;
and step eight, estimating the flow velocity value of the water flow.
The implementation of the embodiment can be realized: the method has the advantages that the method plays a role in more universal water body environment and weaker illumination environment, is easy to use in the landing of hydrologic monitoring intelligent engineering, can automatically mark natural tracers in water flow, namely can generate a classification data set at low cost, and is very valuable in measuring stability and energy conservation and emission reduction of flow rate measurement at night because the linear distance between shooting equipment and the river surface is often more than tens of meters, and the conventional image flow rate measurement method requires illumination to achieve the effect that the river surface is similar to the daytime.
Example 2
As shown in fig. 1-7, a water flow velocity measurement method based on motion enhancement features comprises the following steps:
firstly, acquiring and framing water flow video data, and acquiring water flow surface pictures at the moment t, the moment t+Deltat and the moment t+2Deltat;
and secondly, respectively calculating a frame difference diagram of T1 and a frame difference diagram of T2 by using a frame difference method, wherein T1 is between T time and t+Deltat time, T2 is between t+Deltat time and t+2 Deltat time, and Deltat is the reciprocal of the frame rate of a camera in actual use, such as video of 25 frames per second, and Deltat is 40ms. As shown in the first figure, the left half of the image is an image at the time t, and the right half of the image is an image at the time t+Deltat;
step three, counting disturbance noise, carrying out frame difference processing on a water flow video, counting the fluctuation range of a non-zero value of the water flow video, drawing a distribution diagram, taking the highest point as a normal distribution peak, carrying out noise point statistics on the left half, marking the right side as a maximum noise threshold, fitting a normal distribution curve, wherein the fitted normal distribution curve is shown in figure 2;
step four, filtering disturbance noise, denoising the frame difference image according to the self-adaptive threshold calculated in the step three on the frame difference image, filtering random noise which is actually smaller than the self-adaptive threshold, wherein the random noise is shown in fig. 3, clearing the random noise, and the filtered frame difference image is shown in fig. 4;
step five, the filtered picture is enhanced by utilizing a histogram equalization method, and the enhanced frame difference picture is subjected to secondary difference processing, so that a flow velocity characteristic picture with enhanced motion significance characteristics can be obtained, as shown in fig. 5;
step six, generating water flow characteristic displacement, as shown in fig. 6, on a frame difference diagram with enhanced motion significance, using a common local characteristic description operator to find out characteristic points, and using a classification model to verify whether a reference object corresponding to the characteristic points is reliable or not;
step seven, dynamically filtering outliers of the water flow vector, wherein the outliers are removed according to statistics of results in a period of time because the water flow has certain disorder and randomness, vector points with larger errors are possibly generated by single measurement, and points, which are out of the range of +/-10%, of the highest point in a distribution diagram in the measured value are discarded as shown in fig. 7;
step eight, estimating the flow velocity value of the water flow, wherein as shown in fig. 8, which is a screenshot of a dynamic graph, the physical flow velocity of the water flow after conversion is shown, and the flow velocity and the flow vector are displayed in real time on the left part of the graph;
in this embodiment, the sixth specific step is that, firstly, a SURF operator is selected to perform angular point search, the feature descriptors of all points are traversed by using a nearest neighbor matching algorithm on the found angular points, preliminary matching points are found, then partial disordered mismatching points are filtered by using front-back water flow direction consistency and a flow velocity threshold as constraint conditions, finally, a random sampling consistency algorithm RANSAC is used to determine matching feature key points, namely, all points which can be matched only want to be very close to one displacement vector, and the vector mean value of the feature points is the water flow motion vector.
In the embodiment, the feature point in the step six is determined whether to be a reliable tracer through the tracer classification model during selection, and if not, the feature point is not selected.
In this embodiment, Δt in the first step is the reciprocal of the camera frame rate.
In the present embodiment, the formula for fitting the normal distribution curve in the third step is
Where d is the frame difference, g (d) represents a gaussian fitted to the frame difference histogram, the standard deviation thereof, and a is the statistical height of the histogram, in practice, the noise with highest occurrence frequency.
In the embodiment, in the fifth step, it is assumed that the flow velocity is inconvenient in a small time interval, and for the two-to-two difference images of three continuous images, the motion saliency features have a greater degree of consistency, and a processing thought of non-maximum suppression is introduced, so that the most obvious motion features can be highlighted.
In the embodiment, the tracing classification model is based on a general shallow network and trains a cross-water area flow velocity estimation natural tracer classification data set based on deep learning, wherein a data set positive sample comprises natural floaters generated by flowing and floaters carried by water bodies; negative examples include non-physical shadow movement effects and self-shifting creatures.
The implementation of the embodiment can be realized: the method collects the data set to train the AI model for classification to select the characteristic points, and reduces the cost.
Example 3
As shown in fig. 1-9, a water flow velocity measurement method based on motion enhancement features can automatically mark natural tracers in water flow, namely, a classification data set can be generated at low cost. Fig. 8 is a leaf diagram of the natural tracer in water flow, and fig. 9 is a water pattern diagram of the natural tracer in water flow.
In this embodiment, the natural tracer classification dataset is estimated based on the deep learning of the cross-water flow rate: positive samples include natural floats such as water flowers, eddies, bubbles and the like generated by the flow, and floats such as silt and dry branch leaves carried by the water body; negative examples include non-physical shadow movement effects such as windy tree shadow spots, night lights, and self-variable speed creatures such as fish, birds, insects, etc. in the picture.
In this embodiment, the tracer classification model: the classification model is trained on a generic shallow network, such as MobileNetV 2.
In this embodiment, a flow rate estimation algorithm with better generalization is obtained: and D, using the modified model for flow velocity vector feature point selection optimization in the step six, selecting an object image corresponding to the feature point, inputting a tracer classification model, checking whether the tracer is a reliable tracer, and if not, not selecting the feature point.
It can be realized by this example that in this scenario, the result of using other contact flowmeters is 0.65m/s, while the result of this method fluctuates in the range of 0.63-0.71. The test environments with different water bodies and different illumination are replaced, and the error is about 5% overall.
The foregoing has shown and described the basic principles and main features of the present invention and the advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. The water flow velocity measurement method based on the motion enhancement features is characterized by comprising the following steps of:
step one, acquiring and framing water flow video data to acquire t moment and t+ delta
Water flow surface pictures at times t and t+2 delta t;
step two, using a frame difference method to respectively calculate a frame difference diagram of T1 and a frame difference diagram of T2, wherein T1 is between T time and t+Deltat time, and T2 is between t+Deltat time and t+2 Deltat time;
thirdly, counting disturbance noise, carrying out frame difference processing on the water flow video, counting the fluctuation range of non-zero values of the water flow video, drawing a distribution diagram, taking the highest point as a normal distribution peak, carrying out noise point statistics on the left half side, marking the right side as a maximum noise threshold, and fitting a normal distribution curve;
step four, filtering disturbance noise, denoising the frame difference image according to the self-adaptive threshold calculated in the step three on the frame difference image, filtering random noise which is actually smaller than the self-adaptive threshold, and resetting the random noise;
step five, enhancing the filtered picture by using a histogram equalization method, and performing secondary differential processing on the enhanced frame difference picture to obtain a flow velocity characteristic picture with enhanced motion significance characteristics;
step six, generating water flow characteristic displacement, finding out characteristic points by using common local characteristic description operators on a frame difference diagram with enhanced motion significance, and verifying whether a reference object corresponding to the characteristic points is reliable or not by using a classification model;
step seven, dynamically filtering outliers of the water flow vectors, and discarding points out of a range of +/-10% of the highest point in the distribution diagram in the measured value;
and step eight, estimating the flow velocity value of the water flow.
2. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 1, wherein: the sixth specific step is that firstly, a SURF operator is selected to search angular points, the found angular points are traversed through feature descriptors of all points by using a nearest neighbor matching algorithm, preliminary matching points are found, then partial disordered mismatching points are filtered by using front-back water flow direction consistency and a flow speed threshold value as constraint conditions, finally, a random sampling consistency algorithm RANSAC is used for determining matched feature key points, and vector average values of the feature points are water flow motion vectors.
3. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 1, wherein: and step six, judging whether the characteristic points are reliable tracers or not through a tracer classification model when the characteristic points are selected, and if not, not selecting the characteristic points.
4. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 1, wherein: and delta t in the first step is the reciprocal of the frame rate of the camera.
5. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 1, wherein: the formula for fitting the normal distribution curve in the third step is as follows
Where d is the frame difference, g (d) represents a gaussian fit to the frame difference histogram, σ is its standard deviation, a is the statistical height of the histogram, and in practice is the noise with the highest occurrence frequency.
6. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 1, wherein: in the fifth step, assuming that the flow rate is inconvenient in a time interval, for every two different images of three continuous images, the motion significance features are consistent, and a processing thought of non-maximum suppression is introduced, so that the most obvious motion features can be highlighted.
7. A method of measuring velocity of water flow based on motion enhancement features as claimed in claim 3, wherein: the tracing classification model is based on a general shallow network and trains a cross-water area flow velocity estimation natural tracer classification data set based on deep learning, wherein a data set positive sample comprises natural floaters which are generated by flowing and floaters carried by a water body; negative examples include non-physical shadow movement effects and self-shifting creatures.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111157928.7A CN113822909B (en) | 2021-09-30 | 2021-09-30 | Water flow velocity measurement method based on motion enhancement features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111157928.7A CN113822909B (en) | 2021-09-30 | 2021-09-30 | Water flow velocity measurement method based on motion enhancement features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113822909A CN113822909A (en) | 2021-12-21 |
| CN113822909B true CN113822909B (en) | 2023-12-08 |
Family
ID=78919832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111157928.7A Active CN113822909B (en) | 2021-09-30 | 2021-09-30 | Water flow velocity measurement method based on motion enhancement features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113822909B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106841673A (en) * | 2017-01-16 | 2017-06-13 | 甘肃农业大学 | A kind of river surface mean flow rate measurement apparatus and method |
| CN111160210A (en) * | 2019-12-24 | 2020-05-15 | 天津天地伟业机器人技术有限公司 | Video-based water flow velocity detection method and system |
| CN111275752A (en) * | 2020-01-22 | 2020-06-12 | 中国农业科学院农业信息研究所 | Water flow velocity measurement method and device, computer equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8116527B2 (en) * | 2009-10-07 | 2012-02-14 | The United States Of America As Represented By The Secretary Of The Army | Using video-based imagery for automated detection, tracking, and counting of moving objects, in particular those objects having image characteristics similar to background |
-
2021
- 2021-09-30 CN CN202111157928.7A patent/CN113822909B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106841673A (en) * | 2017-01-16 | 2017-06-13 | 甘肃农业大学 | A kind of river surface mean flow rate measurement apparatus and method |
| CN111160210A (en) * | 2019-12-24 | 2020-05-15 | 天津天地伟业机器人技术有限公司 | Video-based water flow velocity detection method and system |
| CN111275752A (en) * | 2020-01-22 | 2020-06-12 | 中国农业科学院农业信息研究所 | Water flow velocity measurement method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113822909A (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101996992B1 (en) | Apparatus and Method for Measuring Flow Velocity of River using Optical Flow Image Processing | |
| CN111798386B (en) | River channel flow velocity measurement method based on edge identification and maximum sequence density estimation | |
| CN102324016B (en) | Statistical method for high-density crowd flow | |
| CN103218816B (en) | A kind of crowd density estimation method and people flow rate statistical method based on video analysis | |
| CN102564508B (en) | Method for implementing online tests of stream flow based on video images | |
| CN101739551B (en) | Method and system for identifying moving objects | |
| CN106023245B (en) | Moving target detecting method under the static background measured based on middle intelligence collection similarity | |
| Zhao et al. | An improvement of the Space-Time Image Velocimetry combined with a new denoising method for estimating river discharge | |
| Lin et al. | Applications of image recognition for real-time water level and surface velocity | |
| CN107631782B (en) | Water level detection method based on Harris angular point detection | |
| US20220128358A1 (en) | Smart Sensor Based System and Method for Automatic Measurement of Water Level and Water Flow Velocity and Prediction | |
| CN110187142A (en) | Flow monitoring method and system | |
| CN111914695B (en) | Tidal bore monitoring method based on machine vision | |
| Pawaskar et al. | Detection of moving object based on background subtraction | |
| Daigle et al. | A methodology based on Particle image velocimetry for river ice velocity measurement | |
| CN103489012A (en) | Crowd density detecting method and system based on support vector machine | |
| Zhen et al. | IP camera-based LSPIV system for on-line monitoring of river flow | |
| Prakoso et al. | Vehicle detection using background subtraction and clustering algorithms | |
| Zhen et al. | Design and evaluation of an FFT-based space-time image velocimetry (STIV) for time-averaged velocity measurement | |
| Alongi et al. | An automatic ANN-based procedure for detecting optimal image sequences supporting LS-PIV applications for rivers monitoring | |
| Osorio-Cano et al. | A method for extracting surface flow velocities and discharge volumes from video images in laboratory | |
| CN113822909B (en) | Water flow velocity measurement method based on motion enhancement features | |
| Li et al. | A robust filtering algorithm based on the estimation of tracer visibility and stability for large scale particle image velocimetry | |
| Daengsi et al. | Irrigation Management: A Pilot Study for Automatic Water Level Measurement and Report System Development Using Machine Learning Associated with Modified Images | |
| AU2021102276A4 (en) | A Detection Method for Quantity of Residual Plastic Films Based on Heuristic Iterative Algorithm and Its Early Warning System |
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 |