Method for improving accuracy of biometric device of profile motion aquaculture net cage

Abstract

The invention relates to the technical field of cage culture, and provides a method for improving the accuracy of a biometric device of a profile motion culture cage, which comprises the following steps of: acquiring all sonar images and corresponding direction information acquired in a rotation process; feature points are extracted based on the netting contour; filtering the feature points, and generating a sparse point set of the net cage contour by difference values; calculating a body-park radius shape coefficient based on the point set, and performing evaluation after waiting for a plurality of time intervals; the contour feature points are mapped to a three-dimensional coordinate system with sonar as an original point based on the direction angle; and the distance between the net cage contour and the sonar scanning boundary. By evaluating the deformation degree of the net, the method can effectively distinguish whether the state of the netting is good or not in data collected by equipment in one-time profile operation, biomass evaluation can be accurately carried out, if the net is seriously deformed, the equipment can give an alarm and suggest a user to carry out measurement again when the water flow is low, and the measurement efficiency is improved. And evaluation data with relatively low accuracy given by the system due to network deformation interference can be effectively avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of net cage culture, in particular, and particularly relates to a method for improving the accuracy of a biological statistics device for a profile motion net cage. BACKGROUND

[0002] In recent years, the net cage culture industry in China has developed rapidly. Compared with large deep-sea net cages, traditional PE gravity type net cages are more common in China. Understanding the total amount of biological resources in net cage culture has a decisive influence on the process control, user asset assessment, insurance business, and other aspects. Therefore, accurate assessment of the total amount of fish culture in the net cage can bring great convenience to the breeders and insurance business.

[0003] During the initial fish release process, users can accurately understand the number of fish released into the net cage according to the fry counter or manual statistics. However, during the process of marine net cage culture, many factors can cause the death of cultured fish, and sometimes the mortality rate during the entire culture period can exceed 50%. In addition, fish may escape or be stolen during the growth process. Therefore, the total amount of fish in the net cage is dynamically changing. The traditional method of relying on manual experience to assess the total amount of culture in the net cage has a large error, which is not conducive to precise control of feed feeding. Therefore, regular assessment of the total amount of fish culture in the net cage can not only guide precise feeding, but also timely prevent losses.

[0004] Based on CN 116449376A, a profile motion net cage biological statistics device and its working method are disclosed. A profile motion net cage biological statistics device is provided, which uses a mode of sounder winch lifting and propeller rotation to cover the entire net cage. The sounder horizontally rotates to scan and statistically analyze the fish density in different water layers, which conforms to the habit of fish distribution. At the same time, the use of sensors in combination with the propeller can achieve closed-loop control of the scanning direction, resist the rotation of the sonar itself caused by the sea current, and thus avoid missing scanning or repeated scanning in the horizontal direction. Since the sonar system needs to ensure sufficient return frame rate under the restriction of limited bandwidth, the size of the sonar return image is limited. The smaller the range, the higher the resolution of the return image, and the clearer the picture quality, which helps to improve the assessment accuracy. However, if the range is too small, it is easy to cause the local net to exceed the measurement range, resulting in the failure to detect part of the fish population. Therefore, the present application proposes a method for automatically adjusting the scanning range of the device. In addition, experience shows that when affected by the water flow, excessive deformation of the net can cause the deformed net to wrap part of the fish population, making it impossible for the sonar to detect this part of the fish population, which also reduces the accuracy of the biomass assessment. Moreover, when the net cage is severely deformed, the lifting device is also prone to net hanging, which may cause safety accidents. Therefore, a method for automatically assessing the degree of net deformation is proposed. SUMMARY

[0005] To overcome the shortcomings of existing technologies and address the issue of biomass assessment using multibeam sonar in aquaculture cages in both marine and freshwater environments, this invention provides a method for improving the accuracy of biomass assessment using a biostatistical device for profile-motion aquaculture cages.

[0006] This invention is achieved through the following technical solution: a method for improving the accuracy of a biostatistical device for cross-sectional aquaculture cages, specifically including the following steps: Step S1: Control the sonar unit to acquire an image of the cage every 10°-20°, and record the direction and angle corresponding to the image. Step S2: Perform instance segmentation on the acquired image, segment the image, and extract the edge region mask of the netting, i.e., the binarized image; if the netting is not found in the image, enlarge the range and repeat S1 and S2 until the netting is found in the sonar images from all angles. Step S3: Mask the sonar images at each angle using... Morphological erosion is performed, and a contour extraction algorithm is used to extract contours, retaining only the contours with the largest area. A contour point set of the mask is obtained, and the shortest distance between all points in the contour point set and the eroded mask is calculated. Noise points with a distance greater than 10 pixels are removed, completing the point set filtering. The denoised point set is then interpolated to fill in missing parts, obtaining a continuous and dense contour point set. Sampling is performed at set distance intervals; if no actual points are found, linear interpolation is used. Complete the set of points to obtain the sparse net cage outline; Step S4: For the obtained net cage contour point set data from various angles, map it to a three-dimensional spatial coordinate system with the sonar as the origin, according to its corresponding angle, to obtain the three-dimensional contour point set of the net cage. Calculate any point in the point set Horizontal distance from point O = (0,0,0) Vertical distance The sonar scanning range is D, with units of meters. If there exists any point ∃ on the three-dimensional outline of the cage... p i ∈ P , making d h,i ≥ D-1 or d v,i ≥ D-1 ⟹Then the range is determined to be too small; let d max =max( d h,max , d v,max After increasing the range, repeat steps S1, S2, and S3, then execute S4 again, until (2≥) D- d (max>1); Conversely, find the point on the 3D outline of the wire mesh cage that is farthest from the origin in both the horizontal and vertical directions, and calculate (2≥1). D - d max >1) Check if the inequality is true. If it is true, perform the subsequent evaluation. If it is not true, reduce the range to make the inequality true. Step S5: Using the set of mesh cage outline points scanned from various angles, calculate the shape factor SBC based on the Boyce-Clark formula. The shape factor SBC of a single scanning angle is calculated, and the average shape factor of all scanning angles is taken as the basis for judging the deformation of the mesh. Step S6: Based on the Boyce-Clark shape coefficient, with a shape coefficient threshold of 10, determine the degree of deformation of the current cage; if the determination result meets the deformation threshold, proceed with the next step; otherwise, automatically stop the evaluation and wait for a preset time before resuming the entire evaluation process.

[0007] As a preferred option, instance segmentation in step S2 includes Mask-RCNN or YOLO.

[0008] Compared with existing technologies, this invention, by employing the above technical solutions, has the following beneficial effects: For devices that assess fish biomass in net cages using sonar rotation, there are problems such as low data acquisition quality, insufficient counting accuracy, and the inability to automatically adapt measurements to net cages of different sizes due to severe deformation caused by water flow impacts on the netting. This invention proposes a method that can quantitatively calculate the degree of deformation of the net cage caused by water flow without adding additional equipment, and performs sparse three-dimensional reconstruction of the net cage. The three-dimensional reconstruction provides users with an intuitive means to understand the current actual state of the underwater portion of the net cage. Based on the results of the three-dimensional reconstruction, the sonar's range (detection distance) is calculated and automatically adjusted to adapt to the actual size of the net cage. This ensures that the sonar can scan the entire net cage while capturing as many net cages and fish as possible within the limited image size returned by the sonar (since the sonar system needs to ensure a sufficient frame rate under bandwidth limitations, the size of the returned image is limited), improving the detail resolution of the acoustic images and thus effectively improving the accuracy of the system's biomass assessment. By assessing the degree of net deformation, this method can effectively determine whether the net condition is good enough for accurate biomass assessment in the data collected by the equipment in a single profiling operation. If the net is severely deformed, the equipment can issue an alarm and suggest that the user measure again when the water flow is lower, effectively avoiding the interference of net deformation that leads to the system providing less accurate assessment data.

[0009] Additional aspects and advantages of the invention will become apparent in the following description or may be learned by practice of the invention. Attached Figure Description

[0010] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a schematic diagram of the cross-sectional biostatistical device for aquaculture cages. Figure 2 This is a schematic diagram of the process of the present invention; Figure 3 In the middle, (a) is the outline extracted from a normal net cage, and (b) is the outline extracted from a severely deformed net cage; Figure 4 (a) shows the outline extracted from a normal wire mesh cage, and the sparse point set obtained after filtering and interpolation. (b) shows the outline extracted from a severely deformed wire mesh cage, and the sparse point set obtained after filtering and interpolation. Figure 5 (a) shows the sparse point cloud after 3D reconstruction of a normal net cage, and (b) shows the sparse point cloud after 3D reconstruction of a severely deformed net cage.

[0011] Figure 6 (a) is a radial distance diagram of a normal cage scanned at a certain angle, and (b) is a radial distance diagram of a severely deformed cage scanned at a certain angle.

[0012] in, Figure 1 The correspondence between the reference numerals and components in the attached drawings is as follows: 1. Waterproof winch, 2. Winch frame, 3. Sonar unit, 4. Sealed chamber (electrical chamber), 5. Thruster, 6. Sonar, 7. Sonar protective cover, 8. Umbilical cable, 9. Limiting cable, 10. Counterweight, 11. Heading angle sensor, 12. Depth sensor, 13. Lifting cable, 14. Cylindrical frame. Detailed Implementation

[0013] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0014] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0015] The following is combined with Figures 1 to 6 The method for improving the accuracy of the biostatistical device for cross-sectional aquaculture cages according to embodiments of the present invention will be described in detail.

[0016] likeFigure 1 , Figure 2 As shown, this invention proposes a method to improve the accuracy of a biostatistical device for a cross-sectional motion aquaculture cages. Based on the cross-sectional motion aquaculture cage biostatistical device disclosed in CN 116449376A, it includes a waterproof winch 1, with a winch frame 2 mounted on its exterior. The waterproof winch 1 is placed inside the winch frame 2. The top of the winch frame 2 is connected to the cage's railing or steel structure via several hoisting cables 13. An umbilical cable 8 on the waterproof winch 1 is connected to a sonar unit 3, which is pulled by the umbilical cable 8 to perform vertical cross-sectional motion. The sonar unit 3 includes a cylindrical frame 14, with a sealed chamber 4 installed in the upper part of the cylindrical frame 14. Two thrusters 5 are symmetrically and vertically mounted in the middle of the cylindrical frame 14, with opposite rotation directions. The top of the cylindrical frame 14 is connected to the umbilical cable 8, and a sonar protective cover 7 is fixedly installed at the bottom of the cylindrical frame 14. The sonar protective cover 7 contains a sonar 6. The bottom of the waterproof winch 1 is connected to a limit cable 9. The limit cable 9 passes through the sonar unit 3 and the sonar protective cover 7 and is connected to the counterweight 10, which hangs down to the bottom of the cage. The sealed chamber 4 is equipped with a main control circuit board, a depth sensor 12, a heading angle sensor 11, and a thruster drive module. The sonar 6 and the thruster 5 are all connected to the sealed chamber 4 through watertight connectors. The umbilical cable 8 transmits power and data to the sealed chamber 4.

[0017] Specifically, the following steps are included: Step S1: Control the sonar unit to acquire an image of the cage every 10°-20°, and record the direction and angle corresponding to the image. Step S2: Segment the acquired images using instance segmentation algorithms such as Mask-RCNN or YOLO, extracting the edge regions of the mesh (mask, i.e., the binarized image). If the mesh is not found in the image, enlarge the image size and repeat steps S1 and S2 until the mesh is found in sonar images from all angles. The result is as follows. Figure 3 (a), Figure 3 As shown in (b); Step S3: Mask the sonar images at each angle using... Morphological erosion is performed, using common contour extraction algorithms to extract contours, retaining only the contours with the largest area; the contour point set of the mask is obtained, and the shortest distance between all points in the contour point set and the eroded mask is calculated, removing noise points with a distance greater than 10 pixels to complete point set filtering; the denoised point set is interpolated to fill in missing parts, obtaining a continuous and dense contour point set; sampling is performed at set distance intervals, and if no actual points are available, linear interpolation formulas are used. After completion, a sparse set of net cage outline points is finally obtained, as shown in the figure. Figure 4 (a), Figure 4 As shown in (b); Step S4: For the obtained net cage contour point set data from various angles, map it to a three-dimensional spatial coordinate system with the sonar as the origin, according to its corresponding angle, to achieve three-dimensional reconstruction of the entire net cage contour. The result is as follows: Figure 5 (a), Figure 5 As shown in (b), the three-dimensional contour point set of the wire mesh cage is obtained. Calculate any point in the point set The horizontal distance (in the xy plane, horizontal with respect to the ground) from point O=(0,0,0) (the origin of the sonar coordinate system). Vertical distance (distance along the z-axis, perpendicular to the ground) The sonar scanning range is D, with units of meters. If there exists any point ∃ on the three-dimensional outline of the cage... p i ∈ P , making d h,i ≥ D-1 or d v,i ≥ D-1 ⟹Then the range is determined to be too small; let d max =max( d h,max , d v,max After increasing the range, repeat steps S1, S2, and S3, then execute S4 again, until (2≥) D - d (max>1); Conversely, find the point on the 3D outline of the wire mesh cage that is farthest from the origin in both the horizontal and vertical directions, and calculate (2≥1). D - d max >1) Check if the inequality is true. If it is true, perform the subsequent evaluation. If it is not true, reduce the range to make the inequality true, thereby increasing the image resolution before performing the subsequent evaluation. Step S5: Using the set of mesh cage outline points scanned from various angles, calculate the shape factor based on the Boyce-Clark formula. The shape factor (SBC) of a single scanning angle is calculated, and the average of the shape factors of all scanning angles is used as the basis for judging the deformation of the wire mesh. The radial distance map can visualize the differences between images of different wire mesh cages, such as the difference between normal wire mesh cages and severely deformed wire mesh cages. Figure 6 (a), Figure 6 As shown in (b).

[0018] Step S6: Based on the Boyce-Clark shape coefficient, set the shape coefficient threshold to 10 (the threshold can be adjusted by combining measured and simulated data) and judge the degree of deformation of the current cage; if the judgment result meets the deformation threshold, proceed to the next step; otherwise, stop the evaluation automatically and repeat the entire evaluation process after waiting for a preset time.

[0019] In the description of this invention, the term "a plurality of" refers to two or more. Unless otherwise explicitly defined, the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. The terms "connection," "installation," "fixing," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection or an indirect connection through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0020] In the description of this specification, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0021] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for improving the accuracy of a biological statistics device for a profiled aquaculture net cage, characterized by, Specifically comprising the following steps: Step S1, control the sonar unit to collect a net cage image every 10°-20°, and record the direction angle corresponding to the image; Step S2, use instance segmentation on the collected image, segment the image, extract the edge region mask of the netting, that is, the binary image; if no netting is found in the image, after a large amount of dispensing, repeat S1 and S2 until the sonar images of all angles find the netting; Step S3, mask of each angle sonar image, using Line morphological erosion, extract the contour using the contour extraction algorithm, only keep the largest area contour; Get the contour point set of the mask, calculate the shortest distance between all points in the contour point set and the mask after erosion, remove the noise points with a distance greater than 10 pixels, complete the point set filtering; Interpolate the missing part of the denoised point set to obtain a continuous and dense contour point set; Sample at a set distance interval, if there is no actual point, complete it by linear interpolation formula ​ Step S4, mapping the obtained net cage profile point set data of each angle to the three-dimensional coordinate system according to the corresponding angle of the sonar as the origin, and obtaining the net cage three-dimensional profile point set , calculating the horizontal distance of any point in the point set from the point O=(0,0,0) and the vertical distance , the sonar scanning range is D, in meters, if there exists any point ∃ p i ∈ P on the net cage three-dimensional profile, such that d h,i ≥ D-1 or d v,i ≥ D-1 ⟹ then it is determined that the range is too small; let d max =max( d h,max , d v,max )= the net cage three-dimensional profile after the range is enlarged, repeating S1, S2, S3 and then executing S4 again until (2≥ D − d max >1); otherwise, finding the point on the net cage three-dimensional profile that is farthest from the origin in the horizontal and vertical directions, calculating whether (2≥ D − d max >1) is established, if it is established, subsequent evaluation is executed, if it is not established, the range is reduced until the inequality is established Step S5, using the net cage profile point set scanned at each angle, calculating the shape coefficient SBC based on the Boyce-Clark shape coefficient calculation formula SBC= , obtaining the shape coefficient SBC of a single scanning angle, and taking the average of the shape coefficients of all scanning angles as the net deformation judgment basis; Step S6, based on the Boyce-Clark shape coefficient, set the shape coefficient threshold to 10, and judge the deformation degree of the current net cage; If the judgment result meets the deformation threshold, the subsequent steps are executed; otherwise, the evaluation is automatically stopped, and the entire evaluation process is performed again after waiting for a preset time length.

2. A method for improving the accuracy of a biological statistics device for a profiled aquaculture net cage according to claim 1, characterized in that The instance segmentation in the step S2 includes Mask-RCNN or YOLO.

Images