CN116821807B - Machine vision-based fishery object identification and automatic recording method and system - Google Patents
Machine vision-based fishery object identification and automatic recording method and system Download PDFInfo
- Publication number
- CN116821807B CN116821807B CN202311103810.5A CN202311103810A CN116821807B CN 116821807 B CN116821807 B CN 116821807B CN 202311103810 A CN202311103810 A CN 202311103810A CN 116821807 B CN116821807 B CN 116821807B
- Authority
- CN
- China
- Prior art keywords
- information
- fishery
- data
- identification area
- fish catch
- 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
- 238000000034 method Methods 0.000 title claims abstract description 68
- 241000251468 Actinopterygii Species 0.000 claims abstract description 203
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 93
- 238000013145 classification model Methods 0.000 claims abstract description 54
- 244000005700 microbiome Species 0.000 claims abstract description 50
- 239000013598 vector Substances 0.000 claims description 65
- 238000001514 detection method Methods 0.000 claims description 59
- 238000012545 processing Methods 0.000 claims description 33
- 230000008569 process Effects 0.000 claims description 28
- 230000000007 visual effect Effects 0.000 claims description 28
- 230000006870 function Effects 0.000 claims description 25
- 238000012544 monitoring process Methods 0.000 claims description 25
- 238000013500 data storage Methods 0.000 claims description 19
- 238000013507 mapping Methods 0.000 claims description 18
- 238000003708 edge detection Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 15
- 238000004140 cleaning Methods 0.000 claims description 13
- 230000005764 inhibitory process Effects 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 11
- 238000000354 decomposition reaction Methods 0.000 claims description 10
- 238000012876 topography Methods 0.000 claims description 10
- 238000001228 spectrum Methods 0.000 claims description 9
- 238000002835 absorbance Methods 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 4
- 238000010790 dilution Methods 0.000 claims description 4
- 239000012895 dilution Substances 0.000 claims description 4
- 230000008030 elimination Effects 0.000 claims description 4
- 238000003379 elimination reaction Methods 0.000 claims description 4
- 229910052760 oxygen Inorganic materials 0.000 claims description 4
- 239000001301 oxygen Substances 0.000 claims description 4
- 238000011426 transformation method Methods 0.000 claims description 4
- 230000007704 transition Effects 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 241000238424 Crustacea Species 0.000 description 1
- 241000237852 Mollusca Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The invention discloses a machine vision-based method and a machine vision-based system for identifying and automatically recording a fishery object, which comprise the following steps: detecting the water quality condition and the microorganism type and density condition of the identification area to generate first fish catch information; the first fish catch information is combined with the processed sonar information to generate second fish catch information; and the second fish catch information is combined with the processed binary gray level image to generate third fish catch information, the third fish catch information is guided into an SVM classification model to be classified to generate fish catch classification data, a data recording platform is constructed, and the data recording platform automatically records the fish catch classification data.
Description
Technical Field
The invention relates to the field of image recognition, in particular to a machine vision-based fishery object recognition and automatic recording method and system.
Background
Along with the rapid development of fishing technology due to technological progress, the types and the number of the fishery winnings are increasing, and the types and the number of the fishery winnings in the water area are more, so that the types and the number of the fishery winnings need to be identified, counted and managed, and the current method for identifying the types and the number of the fishery winnings is usually to manually conduct classification identification and quantity detection on the fishery winnings, so that manpower and material resources are consumed. The system for identifying and automatically recording the fishery winnings based on machine vision is used for greatly saving manpower and material resources and improving economic benefits, and meanwhile, the supervision department can prevent the fishery winnings from being ensured in the fishing process through identifying and recording the fishery winnings.
Disclosure of Invention
The invention overcomes the defects of the prior art and provides a method and a system for identifying and automatically recording the fishery winnings based on machine vision.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the invention provides a machine vision-based fishery object identification and automatic recording method, which comprises the following steps:
detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
The fish catch classification data are converted into fish catch classification data conforming to the recording format, a data recording platform is constructed, automatic recording is carried out on the fish catch classification data conforming to the recording format, and the data recording platform carries out recording monitoring in the automatic recording process.
Further, in a preferred embodiment of the present invention, the detecting the water quality in the identification area to obtain a water quality detection result, obtaining microorganism type and density data in the identification area by a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and the fishery-win article type history data to obtain first fishery-win article information, which specifically includes:
determining an identification area, and placing a water quality detection sensor in the identification area, wherein the water quality detection sensor is used for collecting water in the identification area to perform water quality detection to generate a water quality monitoring result, and the water quality monitoring result comprises a dissolved oxygen detection result, a temperature detection result and a pH value detection result;
acquiring a water sample of an identification area, and placing the water sample with the detection area into a spectrophotometer, wherein various microorganisms in the water sample absorb light energy with specific wavelength, and the intermediate energy level transition generates a characteristic banded spectrum;
The method comprises the steps of preparing microorganism type-density-absorbance curves through different band spectrums generated by different microorganisms, and reversely deducing microorganism type and density data in an identification area according to the microorganism type-density-absorbance curves;
acquiring historical information of the type of the fishery object in the identification area through big data retrieval;
and generating first fish catch information by combining the water quality monitoring result, the microorganism type and density data in the identification area and the historical information of the fish catch type in the identification area.
Further, in a preferred embodiment of the present invention, a sonar device is placed in the detection area, the sonar device emits a sonar signal and receives a sonar signal reflected in the identification area, analyzes and processes the sonar signal reflected in the identification area, and generates second fishery object information in combination with the first fishery object information, which specifically includes:
transmitting a sonar signal in the identification area through a transmitting device of the sonar equipment, and receiving the sonar signal reflected in the identification area through a receiving device of the sonar equipment;
carrying out wavelet decomposition on the sonar signals reflected in the identification area by using a discrete wavelet transformation method to obtain wavelet coefficients of different frequency readings;
Setting a sonar standard signal threshold, retaining a sonar signal with a wavelet coefficient larger than the sonar standard signal threshold, and removing the sonar signal with the wavelet coefficient smaller than the sonar standard signal threshold;
generating an initial denoising sonar signal by inverse discrete wavelet transformation of the reserved wavelet coefficient, and repeatedly performing wavelet decomposition and noise elimination on the initial denoising sonar signal to generate a denoising sonar signal;
generating fishery object sonar information according to the echo amplitude, the echo amplitude frequency and the echo duration of the denoising processing sonar signal;
and combining the fish catch sonar information with the first fish catch information to obtain second fish catch information.
Further, in a preferred embodiment of the present invention, the acquiring an image captured by an identification area, performing Canny edge detection and binarization processing on the image captured by the identification area, and generating a binarized gray-scale image, specifically includes:
acquiring an identification area shooting image through a camera, and carrying out graying treatment on the identification area shooting image to obtain an identification area graying image;
the identification region graying image is led into a Gaussian filter for Gaussian filtering, and a Gaussian filtering graying image is obtained;
Respectively calculating a minimum width value and a minimum direction of a fishery object in the horizontal direction and the vertical direction of the Gaussian filter gray image by using a Sobel operator, and carrying out non-maximum inhibition on the dilution direction of the edge of the fishery object in the Gaussian filter gray image to obtain a non-maximum inhibition gray image;
presetting a strong edge threshold and a weak edge threshold, acquiring a gray value of the non-maximum inhibition gray image, and screening out pixel points meeting the strong edge threshold and the weak edge threshold from the non-maximum inhibition gray image;
connecting the pixel points meeting the strong edge threshold value with all the complete pixel points with the weak edge threshold value based on a region growing algorithm,
based on the binarization method, pixels on the connection edge are defined as 1, and pixels outside the connection edge are defined as 0, so that a binarized gray image is generated.
Further, in a preferred embodiment of the present invention, the extracting the visual feature information of the fishery object in the binary gray scale image combines the visual feature information of the fishery object with the second fishery object information and introduces the combined information into the SVM classification model for classification processing, so as to generate the fishery object classification data, which specifically includes:
Removing pixel points defined as 0 in the binarized image to obtain a texture feature gray-scale image of the fishery, and obtaining visual feature information of the fishery by the texture feature gray-scale image of the fishery;
combining the visual characteristic information of the fish catch with the second fish catch information to generate third fish catch information;
importing the third fishery object information into an SVM classification model, generating various feature vectors, and performing data cleaning treatment on the feature vectors;
dividing various feature vectors after the result data cleaning treatment into a training set and a testing set, and selecting a Gaussian radial basis function as a kernel function in the SVM classification model;
the Gaussian radial basis function trains an SVM classification model by using a training set, various feature vectors of the training set are mapped to a high-dimensional feature space in the SVM classification model, a hyperplane is constructed by taking the position with the largest spacing distance of the various feature vectors, and the various feature vectors are linearly distributed on the hyperplane;
analyzing various feature vectors on the hyperplane, judging the types and the quantity of the various feature vectors, and carrying out classification processing according to the types of the feature vectors;
evaluating the accuracy, precision, recall rate and F1 score of the trained SVM classification model by using a test set, generating an evaluation result, and correcting the SVM classification model according to the evaluation result;
And (5) deriving an SVM classification model from the feature vectors after the classification processing to obtain the classification data of the fishery.
Further, in a preferred embodiment of the present invention, the converting the classification data of the fishery object into classification data of the fishery object conforming to the recording format, constructing a data recording platform, and automatically recording the classification data of the fishery object conforming to the recording format, where the data recording platform performs recording monitoring during the automatic recording process, specifically:
creating a data text mapping table, editing a fish catch type text in the data text mapping table according to fish catch type historical information, importing the fish catch classification data into the data text mapping table, enabling the fish catch classification data to correspond to the fish catch type text, generating fish catch text information, constructing a data storage library, and storing the fish catch text information into the data storage library;
a data recording platform is constructed, a fish catch classification information recording module is preset in the data recording platform, and the data storage library and the data recording platform are automatically recorded into the fish catch classification information recording module in a full-connection mode;
According to the variety characteristic information of various fish catch, presetting characteristic threshold values of various fish catch varieties, if one fish catch information is not in the range of the characteristic threshold value corresponding to the fish catch variety in the automatic recording process, the fish catch classification information recording module does not record the fish catch information.
The invention also provides a machine vision-based fishery object identification and automatic recording system, which comprises a memory and a processor, wherein the memory stores an identification and automatic recording program, and when the identification and automatic recording program is executed by the processor, the method comprises the following steps:
detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
Acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
the fish catch classification data are converted into fish catch classification data conforming to the recording format, a data recording platform is constructed, automatic recording is carried out on the fish catch classification data conforming to the recording format, and the data recording platform carries out recording monitoring in the automatic recording process.
Further, in a preferred embodiment of the present invention, the converting the classification data of the fishery object into classification data of the fishery object conforming to the recording format, constructing a data recording platform, and automatically recording the classification data of the fishery object conforming to the recording format, where the data recording platform performs recording monitoring during the automatic recording process, specifically:
creating a data text mapping table, editing a fish catch type text in the data text mapping table according to fish catch type historical information, importing the fish catch classification data into the data text mapping table, enabling the fish catch classification data to correspond to the fish catch type text, generating fish catch text information, constructing a data storage library, and storing the fish catch text information into the data storage library;
A data recording platform is constructed, a fish catch classification information recording module is preset in the data recording platform, and the data storage library and the data recording platform are automatically recorded into the fish catch classification information recording module in a full-connection mode;
according to the variety characteristic information of various fish catch, presetting characteristic threshold values of various fish catch varieties, if one fish catch information is not in the range of the characteristic threshold value corresponding to the fish catch variety in the automatic recording process, the fish catch classification information recording module does not record the fish catch information.
The invention solves the technical defects in the background technology, and has the following beneficial effects: detecting the water quality condition and the microorganism type and density condition of the identification area to generate first fish catch information; the first fish catch information is combined with the processed sonar information to generate second fish catch information; and the second fish catch information is combined with the processed binary gray level image to generate third fish catch information, the third fish catch information is guided into an SVM classification model to be classified to generate fish catch classification data, a data recording platform is constructed, and the data recording platform automatically records the fish catch classification data. The invention can identify the types of the fishery gains in water by combining sonar signals, water quality detection and image identification, and classify the fishery gains by using an SVM classifier, so that the condition and the change trend of the fishery resources can be better known, the basis is provided for scientific decision management, and meanwhile, the development condition and the market demand of the fishery economy are evaluated by the fishing quantity different from the cargo, and the basis of the economic decision is provided for fishery practitioners.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other embodiments of the drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a machine vision based method for identifying and automatically recording a fish catch;
FIG. 2 is a flowchart showing a classification process by which third fish information is imported into the SVM classification model;
FIG. 3 shows a program diagram of a machine vision based fish-winning identification and automatic recording system.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description. It should be noted that, in the case of no conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those described herein, and therefore the scope of the present invention is not limited to the specific embodiments disclosed below.
FIG. 1 shows a flow chart of a machine vision based method for identifying and automatically recording a fish catch, comprising the steps of:
s102: detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
s104: a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
s106: acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
s108: extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
s110: the fish catch classification data are converted into fish catch classification data conforming to the recording format, a data recording platform is constructed, automatic recording is carried out on the fish catch classification data conforming to the recording format, and the data recording platform carries out recording monitoring in the automatic recording process.
The fish catch means various aquatic animals, including fish, crustaceans, mollusks and other aquatic animals, which are caught by the fishery activities. In a water area, the type and the number of the fish catch matters are unknown, the fish catch matters in the water area need to be identified and distinguished, and various information of the fish catch matters is recorded on a computer for fishery operators to use.
Further, in a preferred embodiment of the present invention, the detecting the water quality in the identification area to obtain a water quality detection result, obtaining microorganism type and density data in the identification area by a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and the fishery-win article type history data to obtain first fishery-win article information, which specifically includes:
determining an identification area, and placing a water quality detection sensor in the identification area, wherein the water quality detection sensor is used for collecting water in the identification area to perform water quality detection to generate a water quality monitoring result, and the water quality monitoring result comprises a dissolved oxygen detection result, a temperature detection result and a pH value detection result;
acquiring a water sample of an identification area, and placing the water sample with the detection area into a spectrophotometer, wherein various microorganisms in the water sample absorb light energy with specific wavelength, and the intermediate energy level transition generates a characteristic banded spectrum;
The method comprises the steps of preparing microorganism type-density-absorbance curves through different band spectrums generated by different microorganisms, and reversely deducing microorganism type and density data in an identification area according to the microorganism type-density-absorbance curves;
acquiring historical information of the type of the fishery object in the identification area through big data retrieval;
and generating first fish catch information by combining the water quality monitoring result, the microorganism type and density data in the identification area and the historical information of the fish catch type in the identification area.
The water quality detection sensor is used for detecting the water quality in the identification area, and the water quality in the identification area can be obtained to judge the type of the fishery object in the identification area approximately due to different living environments of different types of the fishery object. The method comprises the steps of detecting the types and the densities of microorganisms in an identification area by using a spectrophotometer, analyzing the band spectrum generated by different microorganisms to obtain the types and the distribution densities of the microorganisms, and finally obtaining first fish catch information by combining historical data in the identification area, wherein the first fish catch information is the approximate types and the distribution densities of the fish catches in the identification area. The invention can judge the rough type and distribution density of the fish catch in the identification area through historical data, water quality monitoring data and microorganism type and density data.
Further, in a preferred embodiment of the present invention, a sonar device is placed in the detection area, the sonar device emits a sonar signal and receives a sonar signal reflected in the identification area, analyzes and processes the sonar signal reflected in the identification area, and generates second fishery object information in combination with the first fishery object information, which specifically includes:
transmitting a sonar signal in the identification area through a transmitting device of the sonar equipment, and receiving the sonar signal reflected in the identification area through a receiving device of the sonar equipment;
carrying out wavelet decomposition on the sonar signals reflected in the identification area by using a discrete wavelet transformation method to obtain wavelet coefficients of different frequency readings;
setting a sonar standard signal threshold, retaining a sonar signal with a wavelet coefficient larger than the sonar standard signal threshold, and removing the sonar signal with the wavelet coefficient smaller than the sonar standard signal threshold;
generating an initial denoising sonar signal by inverse discrete wavelet transformation of the reserved wavelet coefficient, and repeatedly performing wavelet decomposition and noise elimination on the initial denoising sonar signal to generate a denoising sonar signal;
generating fishery object sonar information according to the echo amplitude, the echo amplitude frequency and the echo duration of the denoising processing sonar signal;
And combining the fish catch sonar information with the first fish catch information to obtain second fish catch information.
It should be noted that, the frequency of the sonar signal emitted by the emitter of the sonar equipment is 50kHz, and the sonar signal generates an echo when encountering a fishery object in the process of propagating in water, and the echo is a reflected sonar signal. Because echo signals may have signal loss, signal disturbance and the like in the transmission of the echo signals in water, the echo signals need to be processed. The sonar signals smaller than the threshold value of the sonar standard signals in wavelet transformation are noise, the noise needs to be removed, the denoising effect of the signals subjected to primary denoising treatment may be not ideal, the size of the threshold value can be adjusted, different wavelet basis functions can be selected or the number of layers of wavelet decomposition can be adjusted, and the denoising effect is further optimized. And the setting of the threshold value of the sonar standard signal needs to be kept balanced, the signal details are lost due to the fact that the threshold value is too high, and more noise components are reserved due to the fact that the threshold value is too low. And stopping denoising when the denoising treatment enables the echo signal to reach a set stopping condition, such as reaching the required decomposition layer number and reaching a specific signal-to-noise ratio. The echo signals after denoising treatment are analyzed, the types and the quantities of fishes can be roughly judged according to the differences of the amplitude, the frequency and the duration of the echo, and the fishery object sonar information is generated. The combination of the fish catch sonar information and the first fish catch information can further determine the fish catch type in the identification area, and generate second fish catch information. The invention can further determine the kind of the fishery object through sonar signals.
Further, in a preferred embodiment of the present invention, the acquiring an image captured by an identification area, performing Canny edge detection and binarization processing on the image captured by the identification area, and generating a binarized gray-scale image, specifically includes:
acquiring an identification area shooting image through a camera, and carrying out graying treatment on the identification area shooting image to obtain an identification area graying image;
the identification region graying image is led into a Gaussian filter for Gaussian filtering, and a Gaussian filtering graying image is obtained;
respectively calculating a minimum width value and a minimum direction of a fishery object in the horizontal direction and the vertical direction of the Gaussian filter gray image by using a Sobel operator, and carrying out non-maximum inhibition on the dilution direction of the edge of the fishery object in the Gaussian filter gray image to obtain a non-maximum inhibition gray image;
presetting a strong edge threshold and a weak edge threshold, acquiring a gray value of the non-maximum inhibition gray image, and screening out pixel points meeting the strong edge threshold and the weak edge threshold from the non-maximum inhibition gray image;
connecting the pixel points meeting the strong edge threshold value with all the complete pixel points with the weak edge threshold value based on a region growing algorithm,
Based on the binarization method, pixels on the connection edge are defined as 1, and pixels outside the connection edge are defined as 0, so that a binarized gray image is generated.
The aim of the gray processing of the image shot by the identification area is to reduce the image memory, improve the image definition, more intuitively reflect the light and shade changes of the image and enhance the image contrast. The image needs to be filtered due to noise and partial pixel loss. Filtering the gray image of the identification area by using a Gaussian filter, calculating a filter weight matrix, aligning the center point of the filter with the pixels in the current image, carrying out weighted average on the pixels covered by the filter, and summing the products of the filter weight matrix and the gray values of the pixels to realize the filtering.
In addition, the Canny edge detection algorithm is used for searching the position of the fishery object in the image and describing the characteristics of the fishery object. The Sobel operator can position the existence position of the fishery object in the image and judge the size of the fishery object, and the aim of non-maximum inhibition of the dilution direction of the edge of the fishery object in the Gaussian filter gray-scale image is to refine the edge of the fishery object and inhibit the response of the non-edge. The strong edge is the outline edge of the fish catch, and the weak edge is the texture feature, the grain feature and other information of the fish catch. The strong edge and all weak edge pixel points are connected to form a continuous edge, so that the detailed distribution position, shape, texture characteristics and other information of the fishery object in the image can be obtained. And assigning values by using the edges of the fishery gains in the binarization method image to generate a binarization gray level image. The invention can obtain detailed distribution position, shape, texture characteristics and other information of the fishery object in the image through a Canny edge detection algorithm.
FIG. 2 shows a flow chart of a classification process by which third fish information is imported into an SVM classification model, comprising the steps of:
s202: acquiring third fish catch information, importing the third fish catch information into an SVM classification model to obtain various feature vectors, and cleaning data of the various feature vectors;
s204: training the training set of the feature vectors by using an SVM classification model;
s206: and evaluating and correcting the trained SVM classification model.
Further, in a preferred embodiment of the present invention, the obtaining the third fish information, importing the third fish information into an SVM classification model to obtain various feature vectors, and performing data cleaning on the various feature vectors, specifically:
removing pixel points defined as 0 in the binarized image to obtain a texture feature gray-scale image of the fishery, and obtaining visual feature information of the fishery by the texture feature gray-scale image of the fishery;
combining the visual characteristic information of the fish catch with the second fish catch information to generate third fish catch information;
the third fish-winning object information is imported into an SVM classification model to generate various feature vectors;
processing the missing value of the feature vector, and filling the missing value by deleting the feature vector sample containing the missing value and an interpolation method;
Detecting and deleting a repeated value in the feature vector by using a data deduplication method, analyzing outlier data, deleting the outlier data if the number of the outlier data is smaller than a preset value, and using logarithmic transformation on the outlier data if the number of the outlier data is larger than the preset value;
and carrying out normalization processing on the feature vector after data cleaning.
The purpose of eliminating the pixel point defined as 0 in binarization is to extract the texture feature of the fish catch, and generate the fish catch visual feature information, which includes the information of the shape size of the fish catch, the pattern of the fish catch, the texture grain, the number of the fish catch, and the like. The fish catch visual characteristic information is combined with the second fish catch information to further determine the type and the number of the fish catches. The SVM classification model is arranged in an SVM classifier and is used for classifying the third fish catch information according to the types and the number of the fish catches. The third fishery information becomes various feature vectors conforming to the working format of the SVM classification model after being imported into the SVM classification model, and data cleaning is needed for various feature connection because format errors easily occur when the format is changed. Missing values and duplicate values exist in various feature vectors, which need to be processed, and data consistency is maintained. Outlier data is generally defined as noise, affecting feature vector accuracy, and logarithmic transformation of outlier data with an number of outliers greater than a preset value is performed to prevent data information from being deleted too much. The data cleaning is an iterative process, and the data cleaning is repeatedly performed according to the actual situation and the data analysis requirement. The invention can obtain various feature vectors and clean data by obtaining the third fish-winning information.
Further, in a preferred embodiment of the present invention, the training set of feature vectors is trained by the SVM classification model, specifically:
dividing various feature vectors after the result data cleaning treatment into a training set and a testing set, and selecting a Gaussian radial basis function as a kernel function in the SVM classification model;
the Gaussian radial basis function trains an SVM classification model by using a training set, various feature vectors of the training set are mapped to a high-dimensional feature space in the SVM classification model, a hyperplane is constructed by taking the position with the largest spacing distance of the various feature vectors, and the various feature vectors are linearly distributed on the hyperplane;
analyzing various feature vectors on the hyperplane, judging the types and the quantity of the various feature vectors, and carrying out classification processing according to the types of the feature vectors;
the gaussian radial basis function is a kernel function, and functions to map various feature vectors into a high-dimensional feature space by calculating distances between the various feature vectors and the center vector and attenuating the distances in an exponential form. The high-dimensional feature space function is to enable various feature vectors to be distributed linearly, gaussian radial basis function application is achieved through inner product calculation, and the function of accurately classifying the fishery is achieved by defining proper center connection to capture similarity and nonlinear relations among various feature vectors. The hyperplane is a plane in the high-dimensional feature space and is perpendicular to the normal vector, and various feature vectors are linearly distributed on the hyperplane. The classification processing is carried out on various feature vectors on the hyperplane, so that the types of the fishery winnings can be accurately distinguished and the quantity of various fishery winnings can be obtained. The invention can classify various feature vectors through Gaussian radial basis functions.
Further, in a preferred embodiment of the present invention, the evaluating and correcting the trained SVM classification model is specifically:
evaluating the performance of the trained SVM classification model by using evaluation indexes, and generating a model evaluation result, wherein the evaluation indexes comprise accuracy, precision, recall and F1 score;
if the evaluation result of the performance of the trained SVM classification model does not meet the evaluation index requirement, correcting the trained SVM classification model;
analyzing the trained SVM classification model, generating a correction data set, carrying out data preprocessing on the correction data set, adjusting a kernel function in the correction data set, and retraining.
Since the SVM classification model is used for classifying various fishery products, the trained SVM classification model needs to be evaluated and corrected in order to prevent errors in the classification process. The accuracy rate represents the proportion of the feature vector with correct classification to the total feature vector, the accuracy rate represents the proportion of the feature vector with correct classification to the true positive class, the recall rate represents the proportion of the feature vector with correct classification to the true positive class, the F1 score is the harmonic average value of the accuracy rate and the recall rate, and the calculation formula is F1 score = 2 (accuracy rate)/(accuracy rate + recall rate). The kernel function in the modified dataset is adjusted and the cross-validation method can be used to search for the best combination of parameters within a given sum function. The method can evaluate and correct the trained SVM classification model.
In addition, the machine vision-based method for identifying and automatically recording the fish gains further comprises the following steps:
performing laser topography measurement on the identification area by using an unmanned aerial vehicle, transmitting laser beams by using a laser detector on the unmanned aerial vehicle, and obtaining the water depths of all places of the identification area according to the time difference of the laser beams;
constructing a three-dimensional topography model according to the water depths of all places of the identification area, obtaining the living environment of the fishery winnings in the identification area, and obtaining the specific distribution positions of various fishery winnings in the identification area according to the three-dimensional topography model and the living environment of the fishery winnings;
and continuously extracting the characteristics of the fishing and catching tool by using a Canny edge detection method on the Gaussian filtered gray image, and if the Gaussian filtered gray image identifies that the fishing and catching tool exists in the fishing and catching time of the unofficial personnel, making a corresponding scheme.
It should be noted that, the laser beam is by infrared laser beam and green laser beam, the unable surface of water that pierces through of infrared laser beam, direct original road returns unmanned aerial vehicle, and green laser beam can pierce through the surface of water and reach the bottom of water, returns unmanned aerial vehicle again, can obtain the water depth data of corresponding incident point according to the time difference of two kinds of laser beams. And analyzing the Gaussian filtered graying image, and detecting the fishing tool in the identification area by using a Canny edge detection method. The invention can judge the distribution position of the fishery object by constructing a three-dimensional topography model.
As shown in fig. 3, the second aspect of the present invention further provides a program diagram of a machine vision-based automatic and fisher-object identification system, where the automatic and fisher-object identification system includes a memory 31 and a processor 32, and the memory 31 stores an automatic and fisher-object identification program, and when the automatic and fisher-object identification program is executed by the processor 32, the following steps are implemented:
detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
The fish catch classification data are converted into fish catch classification data conforming to the recording format, a data recording platform is constructed, automatic recording is carried out on the fish catch classification data conforming to the recording format, and the data recording platform carries out recording monitoring in the automatic recording process.
Further, in a preferred embodiment of the present invention, the converting the classification data of the fishery object into classification data of the fishery object conforming to the recording format, constructing a data recording platform, and automatically recording the classification data of the fishery object conforming to the recording format, where the data recording platform performs recording monitoring during the automatic recording process, specifically:
creating a data text mapping table, editing a fish catch type text in the data text mapping table according to fish catch type historical information, importing the fish catch classification data into the data text mapping table, enabling the fish catch classification data to correspond to the fish catch type text, generating fish catch text information, constructing a data storage library, and storing the fish catch text information into the data storage library;
a data recording platform is constructed, a fish catch classification information recording module is preset in the data recording platform, and the data storage library and the data recording platform are automatically recorded into the fish catch classification information recording module in a full-connection mode;
According to the variety characteristic information of various fish catch, presetting characteristic threshold values of various fish catch varieties, if one fish catch information is not in the range of the characteristic threshold value corresponding to the fish catch variety in the automatic recording process, the fish catch classification information recording module does not record the fish catch information.
It should be noted that, the text information of all the types of the fish catch that may exist in the identification area is edited on the data text mapping table, the fish catch classification data is imported into the data text mapping table, and the fish catch classification data automatically corresponds to the text information of the corresponding fish catch type, and has a distinguishing function and is stored in the data storage library. The data storage library is associated with the data recording platform, the data in the data storage library is shared with the data recording platform, and the function of the fish catch classification information recording module is to extract and store fish catch classification data in the data storage library. Because of the ease of careless mistakes during the automatic recording process, one type of fish catch is easy to record another similar type of fish catch during the recording process, and therefore monitoring is required during the recording process of fish catch information. The invention can automatically record the fish catch information through the data text mapping table and the fish catch classification information recording module.
The foregoing is merely illustrative embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think about variations or substitutions within the technical scope of the present invention, and the invention should be covered. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (3)
1. The machine vision-based fish-winning object identification and automatic recording method is characterized by comprising the following steps:
detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
Extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
converting the fish catch classification data into fish catch classification data conforming to a recording format, constructing a data recording platform, automatically recording the fish catch classification data conforming to the recording format, and performing recording monitoring in the automatic recording process by the data recording platform;
the method comprises the steps of detecting water quality in an identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish catch type historical data to obtain first fish catch information, wherein the first fish catch information comprises the following specific steps:
determining an identification area, and placing a water quality detection sensor in the identification area, wherein the water quality detection sensor is used for collecting water in the identification area to perform water quality detection to generate a water quality monitoring result, and the water quality monitoring result comprises a dissolved oxygen detection result, a temperature detection result and a pH value detection result;
Acquiring a water sample of an identification area, and placing the water sample with the detection area into a spectrophotometer, wherein various microorganisms in the water sample absorb light energy with specific wavelength, and the intermediate energy level transition generates a characteristic banded spectrum;
the method comprises the steps of preparing microorganism type-density-absorbance curves through different band spectrums generated by different microorganisms, and reversely deducing microorganism type and density data in an identification area according to the microorganism type-density-absorbance curves;
acquiring historical information of the type of the fishery object in the identification area through big data retrieval;
combining the water quality monitoring result, the microorganism type and density data in the identification area and the historical information of the type of the fishery object in the identification area to generate first fishery object information;
the sonar equipment emits sonar signals and receives the sonar signals reflected in the identification area, analyzes and processes the sonar signals reflected in the identification area, and generates second fish catch information by combining the first fish catch information, wherein the second fish catch information comprises:
transmitting a sonar signal in the identification area through a transmitting device of the sonar equipment, and receiving the sonar signal reflected in the identification area through a receiving device of the sonar equipment;
Carrying out wavelet decomposition on the sonar signals reflected in the identification area by using a discrete wavelet transformation method to obtain wavelet coefficients of different frequency readings;
setting a sonar standard signal threshold, retaining a sonar signal with a wavelet coefficient larger than the sonar standard signal threshold, and removing the sonar signal with the wavelet coefficient smaller than the sonar standard signal threshold;
generating an initial denoising sonar signal by inverse discrete wavelet transformation on the reserved wavelet coefficient, and repeatedly performing wavelet decomposition and noise elimination on the initial denoising sonar signal to generate a denoising sonar signal;
generating fishery object sonar information according to the echo amplitude, the echo amplitude frequency and the echo duration of the denoising processing sonar signal;
the fish catch sonar information is combined with the first fish catch information to obtain second fish catch information;
the method comprises the steps of extracting visual characteristic information of a fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data, wherein the method comprises the following specific steps:
removing the pixel point defined as 0 in the binarized gray level image to obtain a fishery texture feature gray level image, and obtaining visual feature information of the fishery from the fishery texture feature gray level image;
Combining the visual characteristic information of the fish catch with the second fish catch information to generate third fish catch information;
importing the third fishery object information into an SVM classification model, generating various feature vectors, and performing data cleaning treatment on the feature vectors;
dividing various feature vectors after the result data cleaning treatment into a training set and a testing set, and selecting a Gaussian radial basis function as a kernel function in the SVM classification model;
the Gaussian radial basis function trains an SVM classification model by using a training set, various feature vectors of the training set are mapped to a high-dimensional feature space in the SVM classification model, a hyperplane is constructed by taking the position with the largest spacing distance of the various feature vectors, and the various feature vectors are linearly distributed on the hyperplane;
analyzing various feature vectors on the hyperplane, judging the types and the quantity of the various feature vectors, and carrying out classification processing according to the types of the feature vectors;
evaluating the accuracy, precision, recall rate and F1 score of the trained SVM classification model by using a test set, generating an evaluation result, and correcting the SVM classification model according to the evaluation result;
the feature vectors after classification processing are led out of an SVM classification model, and the classification data of the fishery is obtained;
The method comprises the steps of converting the fish catch classification data into fish catch classification data conforming to a recording format, constructing a data recording platform, automatically recording the fish catch classification data conforming to the recording format, and carrying out recording monitoring in the automatic recording process by the data recording platform, wherein the specific steps are as follows:
creating a data text mapping table, editing a fish catch type text in the data text mapping table according to fish catch type historical information, importing the fish catch classification data into the data text mapping table, enabling the fish catch classification data to correspond to the fish catch type text, generating fish catch text information, constructing a data storage library, and storing the fish catch text information into the data storage library;
a data recording platform is constructed, a fish catch classification information recording module is preset in the data recording platform, and the data storage library and the data recording platform are automatically recorded into the fish catch classification information recording module in a full-connection mode;
presetting characteristic thresholds of various types of the fishery winnings according to the type characteristic information of the various types of the fishery winnings, and if the type of the fishery winnings is not in the range of the characteristic threshold corresponding to the type of the fishery winnings in the automatic recording process, not recording the type of the fishery winnings by the fishery winnings classification information recording module;
In addition, the machine vision-based method for identifying and automatically recording the fish gains further comprises the following steps:
performing laser topography measurement on the identification area by using an unmanned aerial vehicle, transmitting laser beams by using a laser detector on the unmanned aerial vehicle, and obtaining the water depths of all places of the identification area according to the time difference of the laser beams;
constructing a three-dimensional topography model according to the water depths of all places of the identification area, obtaining the living environment of the fishery winnings in the identification area, and obtaining the specific distribution positions of various fishery winnings in the identification area according to the three-dimensional topography model and the living environment of the fishery winnings;
and continuously extracting the characteristics of the fishing and obtaining tools by using a Canny edge detection method on the Gaussian filtered gray image, and if the Gaussian filtered gray image identifies that the fishing and obtaining tools exist in the fishing and obtaining time of the unofficial personnel, making a response scheme.
2. The machine vision-based fishery object identification and automatic recording method according to claim 1, wherein the acquiring the identification region photographed image, performing Canny edge detection and binarization processing on the identification region photographed image, and generating a binarized gray-scale image comprises:
Acquiring an identification area shooting image through a camera, and carrying out graying treatment on the identification area shooting image to obtain an identification area graying image;
the identification region graying image is led into a Gaussian filter for Gaussian filtering, and a Gaussian filtering graying image is obtained;
respectively calculating a minimum width value and a minimum direction of a fishery object in the horizontal direction and the vertical direction of the Gaussian filter gray image by using a Sobel operator, and carrying out non-maximum inhibition on the dilution direction of the edge of the fishery object in the Gaussian filter gray image to obtain a non-maximum inhibition gray image;
presetting a strong edge threshold and a weak edge threshold, acquiring a gray value of the non-maximum inhibition gray image, and screening out pixel points meeting the strong edge threshold and the weak edge threshold from the non-maximum inhibition gray image;
connecting the pixel points meeting the strong edge threshold value with all the complete pixel points with the weak edge threshold value based on a region growing algorithm;
based on the binarization method, pixels on the connection edge are defined as 1, and pixels outside the connection edge are defined as 0, so that a binarized gray image is generated.
3. The system for identifying and automatically recording the fisher-earnings based on machine vision is characterized by comprising a memory and a processor, wherein an identification and automatic recording program is stored in the memory, and when the identification and automatic recording program is executed by the processor, the following steps are realized:
Detecting water quality in the identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish-winning object type historical data to obtain first fish-winning object information;
a sonar device is put in the detection area, the sonar device emits a sonar signal and receives the sonar signal reflected in the identification area, the sonar signal reflected in the identification area is analyzed and processed, and second fish catch information is generated by combining the first fish catch information;
acquiring an identification area shooting image, and carrying out Canny edge detection and binarization processing on the identification area shooting image to generate a binarized gray level image;
extracting visual characteristic information of the fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with the second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data;
converting the fish catch classification data into fish catch classification data conforming to a recording format, constructing a data recording platform, automatically recording the fish catch classification data conforming to the recording format, and performing recording monitoring in the automatic recording process by the data recording platform;
The method comprises the steps of detecting water quality in an identification area to obtain a water quality detection result, acquiring microorganism type and density data in the identification area through a spectrophotometer, and combining the water quality detection result, the microorganism type and density data and fish catch type historical data to obtain first fish catch information, wherein the first fish catch information comprises the following specific steps:
determining an identification area, and placing a water quality detection sensor in the identification area, wherein the water quality detection sensor is used for collecting water in the identification area to perform water quality detection to generate a water quality monitoring result, and the water quality monitoring result comprises a dissolved oxygen detection result, a temperature detection result and a pH value detection result;
acquiring a water sample of an identification area, and placing the water sample with the detection area into a spectrophotometer, wherein various microorganisms in the water sample absorb light energy with specific wavelength, and the intermediate energy level transition generates a characteristic banded spectrum;
the method comprises the steps of preparing microorganism type-density-absorbance curves through different band spectrums generated by different microorganisms, and reversely deducing microorganism type and density data in an identification area according to the microorganism type-density-absorbance curves;
Acquiring historical information of the type of the fishery object in the identification area through big data retrieval;
combining the water quality monitoring result, the microorganism type and density data in the identification area and the historical information of the type of the fishery object in the identification area to generate first fishery object information;
the sonar equipment emits sonar signals and receives the sonar signals reflected in the identification area, analyzes and processes the sonar signals reflected in the identification area, and generates second fish catch information by combining the first fish catch information, wherein the second fish catch information comprises:
transmitting a sonar signal in the identification area through a transmitting device of the sonar equipment, and receiving the sonar signal reflected in the identification area through a receiving device of the sonar equipment;
carrying out wavelet decomposition on the sonar signals reflected in the identification area by using a discrete wavelet transformation method to obtain wavelet coefficients of different frequency readings;
setting a sonar standard signal threshold, retaining a sonar signal with a wavelet coefficient larger than the sonar standard signal threshold, and removing the sonar signal with the wavelet coefficient smaller than the sonar standard signal threshold;
generating an initial denoising sonar signal by inverse discrete wavelet transformation on the reserved wavelet coefficient, and repeatedly performing wavelet decomposition and noise elimination on the initial denoising sonar signal to generate a denoising sonar signal;
Generating fishery object sonar information according to the echo amplitude, the echo amplitude frequency and the echo duration of the denoising processing sonar signal;
the fish catch sonar information is combined with the first fish catch information to obtain second fish catch information;
the method comprises the steps of extracting visual characteristic information of a fishery object in the binarized gray level image, combining the visual characteristic information of the fishery object with second fishery object information, and importing the combined information into an SVM classification model for classification processing to generate fishery object classification data, wherein the method comprises the following specific steps:
removing the pixel point defined as 0 in the binarized gray level image to obtain a fishery texture feature gray level image, and obtaining visual feature information of the fishery from the fishery texture feature gray level image;
combining the visual characteristic information of the fish catch with the second fish catch information to generate third fish catch information;
importing the third fishery object information into an SVM classification model, generating various feature vectors, and performing data cleaning treatment on the feature vectors;
dividing various feature vectors after the result data cleaning treatment into a training set and a testing set, and selecting a Gaussian radial basis function as a kernel function in the SVM classification model;
The Gaussian radial basis function trains an SVM classification model by using a training set, various feature vectors of the training set are mapped to a high-dimensional feature space in the SVM classification model, a hyperplane is constructed by taking the position with the largest spacing distance of the various feature vectors, and the various feature vectors are linearly distributed on the hyperplane;
analyzing various feature vectors on the hyperplane, judging the types and the quantity of the various feature vectors, and carrying out classification processing according to the types of the feature vectors;
evaluating the accuracy, precision, recall rate and F1 score of the trained SVM classification model by using a test set, generating an evaluation result, and correcting the SVM classification model according to the evaluation result;
the feature vectors after classification processing are led out of an SVM classification model, and the classification data of the fishery is obtained;
the method comprises the steps of converting the fish catch classification data into fish catch classification data conforming to a recording format, constructing a data recording platform, automatically recording the fish catch classification data conforming to the recording format, and carrying out recording monitoring in the automatic recording process by the data recording platform, wherein the specific steps are as follows:
creating a data text mapping table, editing a fish catch type text in the data text mapping table according to fish catch type historical information, importing the fish catch classification data into the data text mapping table, enabling the fish catch classification data to correspond to the fish catch type text, generating fish catch text information, constructing a data storage library, and storing the fish catch text information into the data storage library;
A data recording platform is constructed, a fish catch classification information recording module is preset in the data recording platform, and the data storage library and the data recording platform are automatically recorded into the fish catch classification information recording module in a full-connection mode;
presetting characteristic thresholds of various types of the fishery winnings according to the type characteristic information of the various types of the fishery winnings, and if the type of the fishery winnings is not in the range of the characteristic threshold corresponding to the type of the fishery winnings in the automatic recording process, not recording the type of the fishery winnings by the fishery winnings classification information recording module;
in addition, the machine vision-based method for identifying and automatically recording the fish gains further comprises the following steps:
performing laser topography measurement on the identification area by using an unmanned aerial vehicle, transmitting laser beams by using a laser detector on the unmanned aerial vehicle, and obtaining the water depths of all places of the identification area according to the time difference of the laser beams;
constructing a three-dimensional topography model according to the water depths of all places of the identification area, obtaining the living environment of the fishery winnings in the identification area, and obtaining the specific distribution positions of various fishery winnings in the identification area according to the three-dimensional topography model and the living environment of the fishery winnings;
And continuously extracting the characteristics of the fishing and obtaining tools by using a Canny edge detection method on the Gaussian filtered gray image, and if the Gaussian filtered gray image identifies that the fishing and obtaining tools exist in the fishing and obtaining time of the unofficial personnel, making a response scheme.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311103810.5A CN116821807B (en) | 2023-08-30 | 2023-08-30 | Machine vision-based fishery object identification and automatic recording method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311103810.5A CN116821807B (en) | 2023-08-30 | 2023-08-30 | Machine vision-based fishery object identification and automatic recording method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116821807A CN116821807A (en) | 2023-09-29 |
| CN116821807B true CN116821807B (en) | 2024-01-09 |
Family
ID=88115350
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311103810.5A Active CN116821807B (en) | 2023-08-30 | 2023-08-30 | Machine vision-based fishery object identification and automatic recording method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116821807B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251055A (en) * | 2016-07-26 | 2016-12-21 | 中国水产科学研究院珠江水产研究所 | A kind of fish pass crosses the Acoustic assessment method of fish effect |
| CN109446902A (en) * | 2018-09-22 | 2019-03-08 | 天津大学 | A kind of marine environment based on unmanned platform and the comprehensive cognitive method of target |
| CN111897350A (en) * | 2020-07-28 | 2020-11-06 | 谈斯聪 | Underwater robot device, and underwater regulation and control management optimization system and method |
| CN113569971A (en) * | 2021-08-02 | 2021-10-29 | 浙江索思科技有限公司 | Image recognition-based catch target classification detection method and system |
| CN113822233A (en) * | 2021-11-22 | 2021-12-21 | 青岛杰瑞工控技术有限公司 | Method and system for tracking abnormal fishes cultured in deep sea |
| CN114998776A (en) * | 2022-04-11 | 2022-09-02 | 浙江海洋大学 | Acoustic and image combined fish target identification and statistics device and method |
| CN115423225A (en) * | 2022-11-07 | 2022-12-02 | 中国水产科学研究院南海水产研究所 | Fishing port operation management method and system based on big data |
| CN115953632A (en) * | 2023-01-16 | 2023-04-11 | 中国水产科学研究院渔业机械仪器研究所 | Fishing catch thing discernment and type device |
| CN116343018A (en) * | 2023-04-24 | 2023-06-27 | 中国水产科学研究院南海水产研究所 | Intelligent fishery fishing identification method, system and medium based on image processing |
| CN116596163A (en) * | 2023-07-05 | 2023-08-15 | 中国水产科学研究院南海水产研究所 | Image detection-based white croaker resource distribution prediction system and method |
| CN116609786A (en) * | 2023-05-22 | 2023-08-18 | 农芯(南京)智慧农业研究院有限公司 | Fish counting method and device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102019217730A1 (en) * | 2019-11-18 | 2021-05-20 | Volkswagen Aktiengesellschaft | Method for operating an operating system in a vehicle and operating system for a vehicle |
| DE102019217733A1 (en) * | 2019-11-18 | 2021-05-20 | Volkswagen Aktiengesellschaft | Method for operating an operating system in a vehicle and operating system for a vehicle |
-
2023
- 2023-08-30 CN CN202311103810.5A patent/CN116821807B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251055A (en) * | 2016-07-26 | 2016-12-21 | 中国水产科学研究院珠江水产研究所 | A kind of fish pass crosses the Acoustic assessment method of fish effect |
| CN109446902A (en) * | 2018-09-22 | 2019-03-08 | 天津大学 | A kind of marine environment based on unmanned platform and the comprehensive cognitive method of target |
| CN111897350A (en) * | 2020-07-28 | 2020-11-06 | 谈斯聪 | Underwater robot device, and underwater regulation and control management optimization system and method |
| CN113569971A (en) * | 2021-08-02 | 2021-10-29 | 浙江索思科技有限公司 | Image recognition-based catch target classification detection method and system |
| CN113822233A (en) * | 2021-11-22 | 2021-12-21 | 青岛杰瑞工控技术有限公司 | Method and system for tracking abnormal fishes cultured in deep sea |
| CN114998776A (en) * | 2022-04-11 | 2022-09-02 | 浙江海洋大学 | Acoustic and image combined fish target identification and statistics device and method |
| CN115423225A (en) * | 2022-11-07 | 2022-12-02 | 中国水产科学研究院南海水产研究所 | Fishing port operation management method and system based on big data |
| CN115953632A (en) * | 2023-01-16 | 2023-04-11 | 中国水产科学研究院渔业机械仪器研究所 | Fishing catch thing discernment and type device |
| CN116343018A (en) * | 2023-04-24 | 2023-06-27 | 中国水产科学研究院南海水产研究所 | Intelligent fishery fishing identification method, system and medium based on image processing |
| CN116609786A (en) * | 2023-05-22 | 2023-08-18 | 农芯(南京)智慧农业研究院有限公司 | Fish counting method and device |
| CN116596163A (en) * | 2023-07-05 | 2023-08-15 | 中国水产科学研究院南海水产研究所 | Image detection-based white croaker resource distribution prediction system and method |
Non-Patent Citations (4)
| Title |
|---|
| 刘元林 等.《人与鱼类》.济南：山东科学技术出版社,2013,第212-215页. * |
| 刘君星 等.《分子生物学仪器与实验技术》.哈尔滨：黑龙江科学技术出版社,2009,第230-245页. * |
| 基于 SVM 的决策融合鱼类识别方法;杜伟东 等;《哈尔滨工程大学学报》;第36卷(第5期);第623-627页 * |
| 基于支持向量机的西北太平洋柔鱼渔场预报模型构建;崔雪森 等;《南方水产科学》;第12卷(第05期);第1-7页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116821807A (en) | 2023-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Stajnko et al. | Modelling apple fruit yield using image analysis for fruit colour, shape and texture | |
| Whittaker et al. | Fruit location in a partially occluded image | |
| CN109255757B (en) | Method for segmenting fruit stem region of grape bunch naturally placed by machine vision | |
| CN106296670B (en) | A kind of Edge detection of infrared image based on the watershed Retinex--Canny operator | |
| CN106815819B (en) | More strategy grain worm visible detection methods | |
| Kim et al. | Detection of pinholes in almonds through x–ray imaging | |
| Munoz-Benavent et al. | Automatic Bluefin Tuna sizing using a stereoscopic vision system | |
| Fu et al. | Kiwifruit yield estimation using image processing by an Android mobile phone | |
| CN116692297A (en) | Dustbin detection method, device, equipment and storage medium based on Internet of things | |
| Marcos et al. | Automated benthic counting of living and non-living components in Ngedarrak Reef, Palau via subsurface underwater video | |
| CN116821807B (en) | Machine vision-based fishery object identification and automatic recording method and system | |
| Laggoune et al. | Tree ring analysis | |
| CN107220972B (en) | A kind of quality of poultry eggs discrimination method based on infrared image | |
| Yadav et al. | An automated image processing method for segmentation and quantification of rust disease in maize leaves | |
| Ye et al. | Cucumber appearance quality detection under complex background based on image processing | |
| Thomas et al. | A graphical automated detection system to locate hardwood log surface defects using high-resolution three-dimensional laser scan data | |
| Setiawan et al. | Shrimp body weight estimation in aquaculture ponds using morphometric features based on underwater image analysis and machine learning approach | |
| Bini et al. | Intelligent agrobots for crop yield estimation using computer vision | |
| Sheng et al. | Fuzzy preprocessing and clustering analysis method of underwater multiple targets in forward looking sonar image for AUV tracking | |
| Huang et al. | High-throughput image analysis framework for fruit detection, localization and measurement from video streams | |
| Prabhu et al. | An orientation independent vision based weight estimation model for alphonso mangoes | |
| Arevalo-Ramirez et al. | Predicting the Elevation of Canopy Occluded Ground Points in Dense Forest Regions | |
| Lee et al. | Sensors i: Color imaging and basics of image processing | |
| Sahoo et al. | Optimized entropy based image segmentation | |
| Byrne et al. | Precise image segmentation for forest inventory |
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 |