CN111990314A - System and method for quantitative observation of fish behaviors - Google Patents
System and method for quantitative observation of fish behaviors Download PDFInfo
- Publication number
- CN111990314A CN111990314A CN202010860592.XA CN202010860592A CN111990314A CN 111990314 A CN111990314 A CN 111990314A CN 202010860592 A CN202010860592 A CN 202010860592A CN 111990314 A CN111990314 A CN 111990314A
- Authority
- CN
- China
- Prior art keywords
- image
- fish
- calibration plate
- short
- moving
- 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.)
- Pending
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; CARE OF BIRDS, FISHES, INSECTS; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K63/00—Receptacles for live fish, e.g. aquaria; Terraria
- A01K63/003—Aquaria; Terraria
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; CARE OF BIRDS, FISHES, INSECTS; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K63/00—Receptacles for live fish, e.g. aquaria; Terraria
- A01K63/003—Aquaria; Terraria
- A01K63/006—Accessories for aquaria or terraria
Abstract
The invention discloses a system and a method for quantitative observation of fish behaviors, wherein the system comprises a calibration unit, a binocular vision observation unit, a rapid storage unit and a behavior analysis unit; the binocular vision observation unit is fixed on an adjusting arm of the camera adjusting device through the damping holder, the observation angle of the camera can be conveniently adjusted by using the damping holder, and the observation angle is sent to the quick storage unit; the rapid storage unit comprises an image compression board card and a network storage server, the image collected by the binocular camera is connected with the image compression board card, and the image is compressed by the image compression board card and stored in the storage server; the behavior analysis unit realizes the initial positioning of the fish in the image by using a motion difference method, determines an optimal observation angle by combining similarity evaluation, and then introduces a shape-based centroid estimation algorithm to determine the gravity center of a moving object, so as to realize the positioning of the moving fish, and the motion behavior quantification of the fish in motion speed, acceleration, water layer distribution and the like can be realized by adopting a method for extracting the centroid.
Description
Technical Field
The invention relates to the field of aquaculture, in particular to a system and a method for quantitative observation of fish behaviors.
Background
Currently, underwater viewing is typically a single camera or camcorder based image acquisition. The existing image acquisition technology can only acquire image information of an angle when observing underwater organisms, a large number of visual blind areas exist, the underwater organisms are various in variety and various in form, and the image information acquired by the traditional image acquisition technology is not enough to acquire complete form characteristic information of the observed organisms, so that the observation result lacks objectivity, and the problem is urgently to be solved.
Disclosure of Invention
The invention aims to provide a system and a method for quantitative observation of fish behaviors to solve the problems in the prior art.
The technical problem solved by the invention can be realized by adopting the following technical scheme:
a system for quantitative observation of fish behaviors comprises a calibration unit, a binocular vision observation unit, a rapid storage unit and a behavior analysis unit;
the calibration unit comprises a positioning track, the positioning track is fixed at the top of the glass water tank through a hanging piece, the positioning track comprises a group of track short shafts and a group of track long shafts which are oppositely arranged, a plurality of first clamping grooves are formed in the track short shafts, two ends of a moving rod are arranged on the track short shafts of the positioning track through the first clamping grooves, a plurality of groups of second clamping grooves are formed in the moving rod, the top end of the moving mechanism is detachably connected with the moving rod through the second clamping grooves, and a calibration plate is fixedly connected with the moving mechanism;
the binocular vision observation unit comprises a binocular camera, the binocular camera is fixed on an adjusting arm of the camera adjusting device through a damping holder, the observation height and the left and right positions of the camera are adjusted through the up-down and left-right movement of the adjusting arm, the observation angle of the camera can be conveniently adjusted by using the damping holder, and the camera is sent to the quick storage unit;
the rapid storage unit comprises an image compression board card and a network storage server, images collected by the binocular camera are connected with the image compression board card through a gigabit network cable, the collection frame rate is adjusted through collection software, and after the collection frame rate is determined, the images are compressed by the image compression board card and stored in the storage server;
the behavior analysis unit realizes the initial positioning of the fish in the image by using a motion difference method, determines an optimal observation angle by combining similarity evaluation, and then introduces a shape-based centroid estimation algorithm to determine the gravity center of a moving object, thereby realizing the positioning of the moving fish, and realizing the quantification of the motion behavior by adopting a method for extracting the centroid for the motion speed, the acceleration, the water layer distribution and the like of the fish.
Furthermore, the storage server has 20 hard disks, wherein 2 SSD and 22 6TB HDD, and the image compression board card realizes image fast storage by using an image compression coding algorithm based on FPGA technology, is used for compressing a sampling image from 12MB to 1MB, and can effectively meet the stable sampling frequency of 120 frames/second of the 4-way binocular camera.
Further, the binocular camera comprises a 2-way long-edge binocular vision camera and a 2-way short-edge binocular vision camera.
Furthermore, the calibration plate adopts a dot calibration plate, the size of the calibration plate is 600mm multiplied by 600mm, the calibration plate is provided with 49 target points which are uniformly arranged, the diameter of the target points is 35.5mm, and the center distance is 70 mm.
Further, the method comprises a sampling method, wherein the sampling method comprises long-edge sampling and short-edge sampling;
the long-edge sampling comprises the following steps:
1) installing long-side binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing a moving rod into a short shaft initial position clamping groove, placing a moving mechanism into a long shaft initial position clamping groove, taking the position at the moment as a sampling starting point, setting a target point at the lower left corner of a calibration plate as a world coordinate system origin, and shooting a 1 st frame image;
3) translating the moving mechanism for 2 times along the long axis direction according to the sequence of the positions of the clamping grooves, and respectively shooting 2 nd to 3 rd frames of images; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 4 th frame of image; the moving mechanism is translated twice in a reverse sequence, and images of 5 th to 6 th frames are shot; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, repeating the process until the last slot clamping position stops, and completing long-edge full-view sampling;
4) repeating the steps until the last card slot position stops, and completing long-edge full-view sampling;
the short side sampling steps are as follows:
1) installing short-edge binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing a moving rod in a long shaft initial position clamping groove and a moving mechanism in a short shaft initial position clamping groove on a calibration rod, taking the position as a sampling starting point, setting a target point at the lower left corner of a calibration plate as a world coordinate system origin, and shooting a 1 st frame of image;
3) translating the moving mechanism for 1 time along the long axis direction according to the position sequence of the clamping grooves, and shooting a 2 nd frame image; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 3 rd frame image; the moving mechanism is translated in reverse sequence for 1 time, and the 4 th frame of image is shot; keeping the position of the moving mechanism, and translating the moving rod along the short axis direction;
4) and repeating the steps until the position of the last card slot is stopped, and finishing the short edge full-view sampling.
Further, after the long side and short side full-view sampling is completed, the method further comprises the following processing:
1) acquiring position information of a calibration plate, and after acquiring the position of the calibration plate, utilizing an operator to segment circles in the area, and finding out the number, the perimeter, the coordinate position and the consistent target point in a calibration plate description file;
2) calculating to obtain the pixel coordinates of the target point; setting an X axis of a world coordinate system to be parallel to a long axis of the fish tank, a Y axis to be vertical to the bottom of the fish tank, a Z axis to be parallel to a short axis of the fish tank, and an origin point to be a lower right target point of an initial position of a calibration plate, and combining the moving distance of the calibration plate and the spatial distribution of the target points to obtain world coordinate system coordinates of the target points;
3) dividing effectively collected target points into a training set and a testing set; optimizing the parameters of the training set of the coordinate system by using a genetic algorithm, and outputting an optimal combination to obtain optimized model parameters after iteration reaches the maximum times;
4) constructing an SVR training model of a coordinate system by using the model parameters determined by the genetic algorithm, and predicting three parameters in a world coordinate system; the target point comprises left and right image pixel coordinates and world coordinate information, and the unit of the image coordinate value is a pixel, namely the pixel coordinate of the target point in the image.
Compared with the prior art, the invention has the beneficial effects that:
1. by adopting the mode of combining a plurality of groups of binocular vision equipment and calibration devices, the visual blind area is eliminated, the behavior information of the fishes is completely acquired, and the observation result has good integrity, objectivity and scientificity.
2. The three-dimensional space coordinate information of the target is reconstructed in a multi-angle image acquisition and analysis mode, and the method has the advantages of non-contact, high precision, wide measurement range and the like, and can well meet the calibration requirement.
Drawings
FIG. 1 is a schematic diagram of a system for quantitative observation of fish behavior according to the present invention.
Fig. 2 is a schematic diagram of a calibration unit according to the present invention.
Fig. 3 is a schematic view of a calibration plate according to the present invention.
Fig. 4 is a schematic position diagram of the binocular vision apparatus and the glass water tank according to the present invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
Referring to fig. 1, fig. 2, fig. 3 and fig. 4, the system for quantitatively observing fish behaviors, which is disclosed by the invention, comprises a calibration unit, a binocular vision observation unit, a fast storage unit and a behavior analysis unit;
the calibration unit comprises a positioning track (1), the positioning track (1) is fixed to the top of the glass water tank (3) through a hanging piece (5), the positioning track (1) comprises a group of track short shafts (6) and a group of track long shafts (7) which are arranged oppositely, a plurality of first clamping grooves (61) are formed in the track short shafts (6), two ends of the movable rod (4) are arranged on the track short shafts (6) of the positioning track (1) through the first clamping grooves (61), a plurality of groups of second clamping grooves (41) are formed in the movable rod (4), the top end of the movable mechanism (2) is detachably connected with the movable rod (4) through the second clamping grooves (41), and the calibration plate (7) is fixedly connected with the movable mechanism (2);
the calibration plate (7) adopts a dot calibration plate, the size of the calibration plate (7) is 600mm multiplied by 600mm, the calibration plate (7) is provided with 49 target points which are uniformly arranged, the diameter of each target point is 35.5mm, and the center distance is 70 mm.
The binocular camera is fixed on the adjusting arm of the camera adjusting device through the damping holder, the observing height and the left and right positions of the camera can be conveniently adjusted by moving the adjusting arm up and down and left and right, the observing angle of the camera can be conveniently adjusted by using the damping holder, the collected images of the camera are connected with the image compression board card through the kilomega network cable, the collection frame rate is adjusted through collection software, and after the collection frame rate is determined, the images are compressed by using the image compression board card and stored in the NAS storage server.
The fast storage system comprises a synchronization/NAS network storage server, an image compression board card and the like. The storage server comprises 20 hard disks, wherein 2 SSD blocks and 22 6TB HDD blocks, an image compression board card realizes the rapid image storage by using an image compression coding algorithm based on the FPGA technology, can compress a sampling image from 12MB to 1MB, and can effectively meet the stable sampling frequency of 120 frames/second of a 4-path camera;
the invention also comprises a sampling method, wherein the sampling method comprises long-edge sampling and short-edge sampling;
the long-edge sampling comprises the following steps:
1) installing long-side binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing the moving rod (4) into the short shaft initial position clamping groove, placing the moving mechanism (2) into the long shaft initial position clamping groove, taking the position at the moment as a sampling starting point, setting a target point at the lower left corner of the calibration plate as a world coordinate system origin, and shooting a frame 1 image;
3) translating the moving mechanism for 2 times along the long axis direction according to the sequence of the positions of the clamping grooves, and respectively shooting 2 nd to 3 rd frames of images; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 4 th frame of image; the moving mechanism is translated twice in a reverse sequence, and images of 5 th to 6 th frames are shot; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, repeating the process until the last slot clamping position stops, and completing long-edge full-view sampling;
4) repeating the steps until the last card slot position stops, and completing long-edge full-view sampling;
the short side sampling steps are as follows:
1) installing short-edge binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing a moving rod (4) into a long shaft initial position clamping groove, placing a moving mechanism (2) into a short shaft initial position clamping groove on a calibration rod, taking the position as a sampling starting point, setting a target point at the lower left corner of a calibration plate as a world coordinate system origin, and shooting a frame 1 image;
3) translating the moving mechanism for 1 time along the long axis direction according to the position sequence of the clamping grooves, and shooting a 2 nd frame image; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 3 rd frame image; the moving mechanism is translated in reverse sequence for 1 time, and the 4 th frame of image is shot; keeping the position of the moving mechanism, and translating the moving rod along the short axis direction;
4) and repeating the steps until the position of the last card slot is stopped, and finishing the short edge full-view sampling.
According to the equipment installation layout and the size of the calibration plate, the upper limit of the distance of the calibration plate which can move from left to right is set to 1200mm, and the upper limit of the distance of the calibration rod which can move from front to back is set to 850 mm; after the above sampling process is completed, there will be 51 pairs of 102 images in the sample library, and the first 47 pairs are selected as training set, the 48 th pair as test set, and the last 3 pairs as evaluation samples. Since each picture has 49 target points (71), 2303 target points (71) are shared in the first 47 images as a training set, 49 target points (71) are shared in the 48 th image as a test set, and 147 target points (71) are shared in the last 3 images as evaluation samples.
The calibration method of the binocular vision equipment comprises an error evaluation process, and the method comprises the following steps:
and obtaining the position information of the calibration board through a find _ calltab () operator carried by HALCON software, after obtaining the position of the calibration board, utilizing the find _ marks _ and _ position () operator to segment circles in the area, and finding out consistent target points (71) of the number, the perimeter, the coordinate position and the like of the circles and in the description file of the calibration board.
The pixel coordinates of the target point (71) can be obtained through the operator operation. Setting the X axis of a world coordinate system to be parallel to the long axis of the fish tank, the Y axis to be vertical to the bottom of the fish tank, the Z axis to be parallel to the short axis of the fish tank, and the origin point to be the lower right target point (71) of the initial position of the calibration plate, and combining the moving distance of the calibration plate and the spatial distribution of the target points (71) to obtain the world coordinate system coordinates of the target points (71). A total effect is collected at 2352 target points (71), wherein the first 2303 points are selected as a training set, and the last 49 points are selected as a testing set.
And (3) optimizing the parameters of the training set of the 3 coordinate systems by using a genetic algorithm, comprehensively considering the operation running speed of the algorithm and the convergence of fitness functions of different calibration models, and uniformly selecting the population scale and the evolution algebra of the 3 groups of optimized samples as 60 and 200. The search ranges for parameters C and g are [1000, 10000], [0.0001, 0.000001], respectively. When the iteration reaches the maximum times, the optimal combination is output, and the optimized model parameters are shown in the following table.
The method combining axial error evaluation and single-point error evaluation is adopted, the MSE function is adopted for the axial error evaluation, and the P function is adopted for the single-point error evaluationMSEThe function is specifically as follows:
in the formula, x ', y ' and z ' are actual space coordinates of the measuring points, xi′、yi′、zi' predicting data for a model of a measurement point, diFor all measurement points the actual spatial coordinates in the x, y or z axis, di' model predicted data at X, Y or the Z-axis for all measurement points.
In addition, since the μm measurement accuracy is already a very good measurement accuracy, and the predicted value based on the training model is not affected by the actual measurement accuracy, the predicted value can reach an infinite number of digits, and therefore, in order to ensure the effectiveness of the evaluation accuracy, the accuracy error is set to three digits after the decimal point.
Aiming at the spatial characteristics of fish motion, the designed multi-angle behavior observation platform inevitably has the problem of repeated positioning, compares and analyzes the motion characteristics of the fish shot at different angles, and determines the optimal observation angle by combining the binocular vision positioning principle and different similarity evaluation algorithms; comprehensively considering the motion characteristics and deformation characteristics of the flexible object, comparing different centroid optimization algorithms, and determining the fish motion centroid extraction principle; the mass center is used as a quantitative target point, and the motion characteristics of the fish, such as the motion direction, the motion speed, the acceleration and the like, are quantized on the basis.
The movement of the fish is taken as a space behavior, and 3-dimensional change of the position exists, and in addition, the fish is taken as a flexible moving object and has behavior characteristics of body movement change and the like. Aiming at the problems, in order to solve the observation visual field limitation of a monocular or binocular camera, a fish positioning and tracking method based on background difference and appearance evaluation is provided by combining a multi-view camera calibration model. The method realizes the preliminary positioning of the moving fish in the 4-angle camera image by using a motion difference method, determines the optimal observation angle by combining a similarity evaluation technology, and then introduces a shape-based centroid estimation algorithm to determine the centroid of the moving object, thereby realizing the positioning of the moving fish, and realizing the quantification of the moving behavior by adopting a method for extracting the centroid for the fish moving speed, acceleration, water layer distribution and the like.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (6)
1. A system for quantitative observation of fish behaviors is characterized by comprising a calibration unit, a binocular vision observation unit, a rapid storage unit and a behavior analysis unit;
the calibration unit comprises a positioning track (1), the positioning track (1) is fixed to the top of the glass water tank (3) through a hanging piece (5), the positioning track (1) comprises a group of track short shafts (6) and a group of track long shafts (7) which are arranged oppositely, a plurality of first clamping grooves (61) are formed in the track short shafts (6), two ends of the movable rod (4) are arranged on the track short shafts (6) of the positioning track (1) through the first clamping grooves (61), a plurality of groups of second clamping grooves (41) are formed in the movable rod (4), the top end of the movable mechanism (2) is detachably connected with the movable rod (4) through the second clamping grooves (41), and the calibration plate (7) is fixedly connected with the movable mechanism (2);
the binocular vision observation unit comprises a binocular camera, the binocular camera is fixed on an adjusting arm of the camera adjusting device through a damping holder, the observation height and the left and right positions of the camera are adjusted through the up-down and left-right movement of the adjusting arm, the observation angle of the camera can be conveniently adjusted by using the damping holder, and the camera is sent to the quick storage unit;
the rapid storage unit comprises an image compression board card and a network storage server, images collected by the binocular camera are connected with the image compression board card through a gigabit network cable, the collection frame rate is adjusted through collection software, and after the collection frame rate is determined, the images are compressed by the image compression board card and stored in the storage server;
the behavior analysis unit realizes the initial positioning of the fish in the image by using a motion difference method, determines an optimal observation angle by combining similarity evaluation, and then introduces a shape-based centroid estimation algorithm to determine the gravity center of a moving object, thereby realizing the positioning of the moving fish, and realizing the quantification of the motion behavior by adopting a method for extracting the centroid for the motion speed, the acceleration, the water layer distribution and the like of the fish.
2. The system for fish behavior quantitative observation according to claim 1, wherein the storage server comprises 20 hard disks, wherein 2 SSD blocks and 22 6TB HDD blocks are provided, and the image compression board card is used for realizing image fast storage by using an image compression coding algorithm based on FPGA technology, is used for compressing a sampled image from 12MB to 1MB, and can effectively meet the stable sampling frequency of 120 frames/second of the 4-way binocular camera.
3. The system for fish behavioral quantification observation according to claim 1, wherein the binocular cameras include a 2-way long-side binocular vision camera and a 2-way short-side binocular vision camera.
4. The system for fish behavior quantification observation according to claim 1, wherein the calibration plate (7) is a circular dot calibration plate, the size of the calibration plate (7) is 600mm x 600mm, the calibration plate (7) has 49 uniformly arranged target points, the diameter of the target points is 35.5mm, and the center-to-center distance is 70 mm.
5. The system for fish behavioral quantification observation according to claim 1, further comprising a sampling method, the sampling method including long-edge sampling and short-edge sampling;
the long-edge sampling comprises the following steps:
1) installing long-side binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing the moving rod (4) into the short shaft initial position clamping groove, placing the moving mechanism (2) into the long shaft initial position clamping groove, taking the position at the moment as a sampling starting point, setting a target point at the lower left corner of the calibration plate as a world coordinate system origin, and shooting a frame 1 image;
3) translating the moving mechanism for 2 times along the long axis direction according to the sequence of the positions of the clamping grooves, and respectively shooting 2 nd to 3 rd frames of images; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 4 th frame of image; the moving mechanism is translated twice in a reverse sequence, and images of 5 th to 6 th frames are shot; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, repeating the process until the last slot clamping position stops, and completing long-edge full-view sampling;
4) repeating the steps until the last card slot position stops, and completing long-edge full-view sampling;
the short side sampling steps are as follows:
1) installing short-edge binocular vision equipment, a calibration plate moving device and a calibration plate, and adjusting the position of the binocular vision equipment to enable the glass water tank to be capable of forming images completely;
2) placing a moving rod (4) into a long shaft initial position clamping groove, placing a moving mechanism (2) into a short shaft initial position clamping groove on a calibration rod, taking the position as a sampling starting point, setting a target point at the lower left corner of a calibration plate as a world coordinate system origin, and shooting a frame 1 image;
3) translating the moving mechanism for 1 time along the long axis direction according to the position sequence of the clamping grooves, and shooting a 2 nd frame image; keeping the position of the moving mechanism, translating the moving rod along the short axis direction, placing the moving rod at the position of the next clamping groove of the positioning track, and shooting a 3 rd frame image; the moving mechanism is translated in reverse sequence for 1 time, and the 4 th frame of image is shot; keeping the position of the moving mechanism, and translating the moving rod along the short axis direction;
4) and repeating the steps until the position of the last card slot is stopped, and finishing the short edge full-view sampling.
6. The system for fish behavioral quantitative observation according to claim 5, wherein after the long-side and short-side full-view sampling is completed, the method further comprises the following steps:
1) acquiring position information of a calibration plate, after the position of the calibration plate is acquired, utilizing an operator to segment circles in the area, and finding out the number, the perimeter and the coordinate position of the circles and a consistent target point (71) in a calibration plate description file;
2) calculating to obtain the pixel coordinates of the target point (71); setting an X axis of a world coordinate system to be parallel to a long axis of the fish tank, a Y axis to be vertical to the bottom of the fish tank, a Z axis to be parallel to a short axis of the fish tank and an origin point to be a lower right target point (71) of the initial position of the calibration plate, and combining the moving distance of the calibration plate and the spatial distribution of the target points (71) to obtain world coordinate system coordinates of the target points (71);
3) dividing effectively collected target points (71) into a training set and a testing set; optimizing the parameters of the training set of the coordinate system by using a genetic algorithm, and outputting an optimal combination to obtain optimized model parameters after iteration reaches the maximum times;
4) constructing an SVR training model of a coordinate system by using the model parameters determined by the genetic algorithm, and predicting three parameters in a world coordinate system; the target point (71) comprises left and right image pixel coordinates and world coordinate information, and the unit of the image coordinate value is a pixel, namely the pixel coordinate of the target point (71) in the image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010860592.XA CN111990314A (en) | 2020-08-25 | 2020-08-25 | System and method for quantitative observation of fish behaviors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010860592.XA CN111990314A (en) | 2020-08-25 | 2020-08-25 | System and method for quantitative observation of fish behaviors |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111990314A true CN111990314A (en) | 2020-11-27 |
Family
ID=73470738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010860592.XA Pending CN111990314A (en) | 2020-08-25 | 2020-08-25 | System and method for quantitative observation of fish behaviors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111990314A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114097696A (en) * | 2021-11-30 | 2022-03-01 | 北京市生态环境保护科学研究院 | Fish quantity counting system and method based on underwater video |
| TWI826231B (en) * | 2023-01-10 | 2023-12-11 | 葉期財 | Container with monitoring device as well as monitoring method thereof |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104036488A (en) * | 2014-05-04 | 2014-09-10 | 北方工业大学 | Binocular vision-based human body posture and action research method |
| CN104680556A (en) * | 2015-02-16 | 2015-06-03 | 燕山大学 | Parallax-based three-dimensional trajectory tracking method of fish movement |
| CN107590466A (en) * | 2017-09-14 | 2018-01-16 | 中国水产科学研究院渔业机械仪器研究所 | A kind of Fish behavior quantization method based on machine vision |
| CN110044374A (en) * | 2018-01-17 | 2019-07-23 | 南京火眼猴信息科技有限公司 | A kind of method and odometer of the monocular vision measurement mileage based on characteristics of image |
| CN111161407A (en) * | 2019-12-26 | 2020-05-15 | 博雅工道（北京）机器人科技有限公司 | Underwater three-dimensional reconstruction system and method |
-
2020
- 2020-08-25 CN CN202010860592.XA patent/CN111990314A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104036488A (en) * | 2014-05-04 | 2014-09-10 | 北方工业大学 | Binocular vision-based human body posture and action research method |
| CN104680556A (en) * | 2015-02-16 | 2015-06-03 | 燕山大学 | Parallax-based three-dimensional trajectory tracking method of fish movement |
| CN107590466A (en) * | 2017-09-14 | 2018-01-16 | 中国水产科学研究院渔业机械仪器研究所 | A kind of Fish behavior quantization method based on machine vision |
| CN110044374A (en) * | 2018-01-17 | 2019-07-23 | 南京火眼猴信息科技有限公司 | A kind of method and odometer of the monocular vision measurement mileage based on characteristics of image |
| CN111161407A (en) * | 2019-12-26 | 2020-05-15 | 博雅工道（北京）机器人科技有限公司 | Underwater three-dimensional reconstruction system and method |
Non-Patent Citations (1)
| Title |
|---|
| 刘世晶等: "基于全视域GA-SVR 模型的鱼类行为双目视觉观测系统标定", 《农业工程学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114097696A (en) * | 2021-11-30 | 2022-03-01 | 北京市生态环境保护科学研究院 | Fish quantity counting system and method based on underwater video |
| TWI826231B (en) * | 2023-01-10 | 2023-12-11 | 葉期財 | Container with monitoring device as well as monitoring method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8803943B2 (en) | Formation apparatus using digital image correlation | |
| CN108107462B (en) | RTK and high-speed camera combined traffic sign post attitude monitoring device and method | |
| CN104537707B (en) | Image space type stereoscopic vision moves real-time measurement system online | |
| CN109272537A (en) | A kind of panorama point cloud registration method based on structure light | |
| JP2020506487A (en) | Apparatus and method for obtaining depth information from a scene | |
| CN111667536A (en) | Parameter calibration method based on zoom camera depth estimation | |
| CN111990314A (en) | System and method for quantitative observation of fish behaviors | |
| CN109859272A (en) | A kind of auto-focusing binocular camera scaling method and device | |
| CN106875436B (en) | It is a kind of based on feature dot density by the method and apparatus of focusing storehouse estimating depth | |
| CN111398274B (en) | Small target object 3D collection equipment | |
| CN109794948A (en) | Distribution network live line work robot and recognition positioning method | |
| CN113470090A (en) | Multi-solid-state laser radar external reference calibration method based on SIFT-SHOT characteristics | |
| CN103198481B (en) | A kind of camera marking method | |
| CN107038753B (en) | Stereoscopic vision three-dimensional reconstruction system and method | |
| CN106709955A (en) | Space coordinate system calibrate system and method based on binocular stereo visual sense | |
| CN107146223B (en) | The analysis system and method for a kind of power transmission tower and power transmission line displacement coupling | |
| CN113393439A (en) | Forging defect detection method based on deep learning | |
| CN108830317B (en) | Rapid and fine evaluation method for joint attitude of surface mine slope rock mass based on digital photogrammetry | |
| CN112465778A (en) | Underwater fish shoal observation device and method | |
| CN105139376A (en) | Shooting counting method based on shooting counting device | |
| CN113446957A (en) | Three-dimensional contour measuring method and device based on neural network calibration and speckle tracking | |
| CN114492070A (en) | High-precision mapping geographic information virtual simulation technology and device | |
| CN110702015B (en) | Method and device for measuring icing thickness of power transmission line | |
| CN114943988B (en) | Planar target monitoring method based on instance segmentation and deep convolution neural network | |
| CN107607737B (en) | Gyro free method for measuring angular velocity based on starlight vector |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201127 |
|
| RJ01 | Rejection of invention patent application after publication |