A method and system for visual monitoring of insect conditions

By using the camera module in the pest monitoring system to extract the characteristics of the insect swarm and use LoRa communication for convolutional identification, the problems of manual inspection error and time-consuming data transmission in pest monitoring are solved, and efficient and accurate generation of pest monitoring results is achieved. .

CN118020736BActive Publication Date: 2024-12-03ANHUI BESTAVI TECH CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202410152403.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-02-03
Publication Date
2024-12-03
Estimated Expiration
2044-02-03

AI Technical Summary

Technical Problem

The existing pest monitoring methods rely on manual inspection and manual photography, and human errors and timeliness are difficult to guarantee. LoRa communication takes a long time to transmit pest visual data, affecting monitoring efficiency.

Method used

After the distribution characteristics and acquisition parameters of the insect attracting point are determined, the image is acquired and the insect swarm characteristics are extracted. The features are uploaded to the general control terminal for convolutional recognition through LoRa communication, and the distribution image is generated as monitoring results.

Benefits of technology

An efficient remote pest visual monitoring system was built. Through distributed architecture and convolutional recognition technology, the transmission efficiency and monitoring accuracy of pest data are improved, and the possibility of human error is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118020736B_ABST
    Figure CN118020736B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of visual monitoring of insect conditions, and specifically discloses a method and system for visual monitoring of insect conditions, the method comprising taking an insect attraction point as a target, determining the distribution characteristics and acquisition parameters of a camera module; obtaining an image of the insect attraction point by a camera module, extracting insect swarm features from the image of the insect attraction point, and uploading the insect swarm features to a master control terminal; the master control terminal performs convolution recognition on the insect swarm features according to a preset convolution kernel to obtain the type and number of insect swarms; the insect swarm features are determined by a histogram of the insect attraction point image; and generating a distribution image according to the type and number of insect swarms at each insect attraction point, as a monitoring result and outputting it. The present invention sets a plurality of acquisition terminals and insect attraction devices, constructs a distributed architecture, acquires visual data by the acquisition terminal and performs feature extraction, and uploads only the extracted features to the master control terminal with the help of a LoRa communication device, and the master control terminal performs feature comparison to obtain an insect condition monitoring result.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of visual monitoring of insect infestation, and in particular to a method and system for visual monitoring of insect infestation. Background Art

[0002] Pests are a common phenomenon in warehouse storage. Existing methods for monitoring pests mostly use attractants to trap pests, and then conduct manual inspections and record and analyze the pest situation by taking photos manually. However, manual inspections are prone to human errors, and the timeliness of detection is difficult to guarantee. In actual scenarios, each trapping point is scattered, and data collection is extremely time-consuming and labor-intensive. To address such issues, a more accurate and convenient monitoring method is needed.

[0003] With the development of communication technology, there are more and more ways of wireless communication. LoRa communication is a communication technology suitable for regional visual monitoring of insect pests. LoRa communication has the remarkable characteristics of low power consumption, long transmission distance, and simple deployment. It can remotely monitor each trapping point with the help of a camera. However, due to the slow communication speed of LoRa communication, it takes a very long time to transmit visual data of insect pests. How to improve the transmission efficiency of visual data is the technical problem that the technical solution of the present invention aims to solve. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for visually monitoring insect conditions to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for visually monitoring insect conditions, the method comprising:

[0007] Taking the insect attracting point as the target, determining the distribution characteristics and acquisition parameters of the camera module; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera;

[0008] The camera module obtains the image of the insect attraction point, extracts the insect swarm features from the image of the insect attraction point, and uploads the insect swarm features to the master control terminal;

[0009] The main control terminal performs convolution identification on the insect swarm features according to a preset convolution kernel to obtain the type and number of insect swarms; the insect swarm features are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method;

[0010] According to the type and number of insect swarms at each insect trapping point, a distribution image is generated and output as the monitoring result.

[0011] As a further solution of the present invention: the step of determining the distribution characteristics and acquisition parameters of the camera module with the insect attraction point as the target includes:

[0012] Obtain the placement location of the insect attractant and query the coordinates of the placement location;

[0013] Counting the coordinates of all placement positions, and determining the installation position and working parameters of the fixed camera points based on the coordinates;

[0014] Query the rated range of the steering camera point, cluster the coordinates according to the rated range, and determine the installation location and working parameters of the steering camera point;

[0015] The process of clustering coordinates according to the rated range is:

[0016] Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates;

[0017] Calculate the center of the same type of coordinates, and based on the center, count other coordinates within the rated range as the same type of coordinates. Execute in a loop until the distance difference between the centers of two adjacent statistical processes is less than a preset value.

[0018] As a further solution of the present invention: the step of acquiring the insect attraction point image by the camera module, extracting the insect swarm features from the insect attraction point image, and uploading the insect swarm features to the master control terminal includes:

[0019] The camera module obtains an image of the insect attraction point; when obtaining the image of the insect attraction point, the time and the coordinates of the insect attraction point are recorded as a label of the image of the insect attraction point;

[0020] Compare the images of the same insect trapping point according to the labels, and determine the monitoring accuracy of each insect trapping point according to the comparison results;

[0021] When the monitoring accuracy is less than a preset accuracy threshold, an error message is generated;

[0022] When the monitoring accuracy is not less than a preset accuracy threshold, the insect swarm features in the insect attraction point image are extracted and uploaded to the master control terminal.

[0023] As a further solution of the present invention: when the monitoring accuracy is not less than a preset accuracy threshold, the step of extracting insect swarm features in the insect attraction point image and uploading it to the master control terminal includes:

[0024] When the monitoring accuracy is not less than a preset accuracy threshold, the background is removed from the insect attraction point image according to a preset background color value;

[0025] Extract three channels of the insect attraction point image after removing the background to obtain three layers;

[0026] Traverse each layer and generate a histogram;

[0027] Calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate the conversion rule;

[0028] The histogram is equalized according to the conversion rule, and three converted histograms are obtained as the swarm characteristics.

[0029] As a further solution of the present invention: the step of performing convolution recognition on the insect swarm characteristics by the master control terminal according to a preset convolution kernel to obtain the type and number of the insect swarm includes:

[0030] Establishing a connection channel with a preset convolution kernel table, and reading convolution kernels in sequence; the convolution kernel table includes convolution kernel items and swarm parameter items; the swarm parameters include swarm type and swarm quantity;

[0031] Calculate the similarity between the convolution kernel and the swarm characteristics, select the convolution kernel with the highest similarity, and read the corresponding swarm parameters;

[0032] The similarity calculation process is:

[0033] In the formula, S is the similarity, exp{-x} is the composite function of the exponential function, is the distance between two histograms, H 1i is the value of the i-th column in the first histogram, H 2i is the value of the i-th column in the second histogram, n is the total number of columns, is the mean value of the first histogram, is the mean of the second histogram.

[0034] As a further solution of the present invention: the step of generating a distribution image according to the type and number of insect swarms at each insect attraction point and outputting the image as a monitoring result comprises:

[0035] Count the type and number of insect swarms at each insect trapping point;

[0036] Query the shape corresponding to the swarm type, and determine the size according to the number of swarms; the relationship between the shape and the swarm type, and the relationship between the size and the number of swarms are pre-set by the staff;

[0037] Based on the statistical shapes and sizes of the insect traps, a distribution image is generated and output as the monitoring result.

[0038] The technical solution of the present invention also provides a visual monitoring system for insect conditions, the system comprising:

[0039] The camera module calibration module is used to determine the distribution characteristics and acquisition parameters of the camera module with the insect attraction point as the target; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera;

[0040] The insect swarm feature extraction module is used to obtain the insect attraction point image through the camera module, extract the insect swarm features from the insect attraction point image, and upload the insect swarm features to the master control terminal;

[0041] A convolution recognition module is used for the master control terminal to perform convolution recognition on insect swarm features according to a preset convolution kernel to obtain the type and number of insect swarms; the insect swarm features are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method;

[0042] The result output module is used to generate a distribution image according to the type and number of insect swarms at each insect trapping point and output it as a monitoring result.

[0043] As a further solution of the present invention: the camera module calibration module includes:

[0044] A coordinate query unit, used to obtain the placement location of the insect attractant and query the coordinates of the placement location;

[0045] A coordinate statistics unit, used to count the coordinates of all placement positions, and determine the installation position and working parameters of the fixed camera point based on the coordinates;

[0046] A clustering unit, used to query the rated range of the steering camera point, cluster the coordinates according to the rated range, and determine the installation position and working parameters of the steering camera point;

[0047] The process of clustering coordinates according to the rated range is:

[0048] Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates;

[0049] Calculate the center of the same type of coordinates, and based on the center, count other coordinates within the rated range as the same type of coordinates. Execute in a loop until the distance difference between the centers of two adjacent statistical processes is less than a preset value.

[0050] As a further solution of the present invention: the insect swarm feature extraction module includes:

[0051] An image acquisition unit is used to acquire an image of an insect attraction point through a camera module; when acquiring the image of the insect attraction point, the time and the coordinates of the insect attraction point are recorded as a label of the image of the insect attraction point;

[0052] An image comparison unit is used to compare images of the same insect attraction point according to the labels, and determine the monitoring accuracy of each insect attraction point according to the comparison results;

[0053] An error reporting unit, configured to generate an error reporting message when the monitoring accuracy is less than a preset accuracy threshold;

[0054] The extraction execution unit is used to extract the insect swarm features in the insect attraction point image and upload it to the master control terminal when the monitoring accuracy is not less than a preset accuracy threshold.

[0055] As a further solution of the present invention: the extraction execution unit includes:

[0056] A rejection subunit, used to reject the background in the insect attraction point image according to a preset background color value when the monitoring accuracy is not less than a preset accuracy threshold;

[0057] The extraction sub-unit is used to extract three channels of the insect attraction point image after removing the background, and obtain three layers;

[0058] Traverse each layer and generate a histogram;

[0059] A rule generation subunit is used to calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate a conversion rule;

[0060] The equalization subunit is used to equalize the histogram according to the conversion rule to obtain three converted histograms as the swarm characteristics.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention sets up multiple collection terminals and insect attracting devices, constructs a distributed architecture, obtains visual data and performs feature extraction at the collection terminal, and uploads only the extracted features to the master control terminal with the help of the LoRa communication device, and the master control terminal performs feature comparison to obtain the insect monitoring results, thereby building an extremely efficient remote pest visual monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0063] Figure 1 This is a flowchart of the visual monitoring method for insect infestation.

[0064] Figure 2 This is the first sub-process flowchart of the insect visual monitoring method.

[0065] Figure 3 This is the second sub-process flowchart of the insect visual monitoring method.

[0066] Figure 4 This is the third sub-process flowchart of the insect visual monitoring method.

[0067] Figure 5 This is the fourth sub-process flowchart of the insect visual monitoring method.

[0068] Figure 6 This is a structural diagram of the insect visual monitoring system. DETAILED DESCRIPTION

[0069] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0070] Figure 1 The flowchart of the method for visual monitoring of insect infestation is shown in the figure. In an embodiment of the present invention, a method for visual monitoring of insect infestation is provided, and the method comprises:

[0071] Step S100: Taking the insect attracting point as the target, determining the distribution characteristics and acquisition parameters of the camera module; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera;

[0072] The insect attraction point is the location where the insect attraction board is installed. In an area, the number of insect attraction boards is limited and is placed in advance by the staff. After the placement is completed, it is necessary to obtain information about the insect attraction boards; considering the timeliness issue, this application uses a camera to collect images of each insect attraction point. The necessary condition for completing this function is that the monitoring area of ​​the camera needs to include the insect attraction point. Therefore, it is necessary to determine the position of each camera and its working parameters with the insect attraction point as the target; the working parameters are the acquisition parameters in the above content, including direction, wide angle and clarity. The direction refers to the direction of the center line of the camera, the wide angle is used to characterize the observation range of the camera, and the clarity is the content richness of the image.

[0073] Step S200: acquiring an image of an insect attraction point by a camera module, extracting insect swarm features from the image of the insect attraction point, and uploading the insect swarm features to a master control terminal;

[0074] After the camera module is installed, the camera module obtains the image of the insect attraction point. In one example of the technical solution of the present invention, a microprocessor (such as a GPU) is also installed on the camera to process the image, so that the data is simplified and then uploaded through the LoRa wireless communication technology. Since the LoRa wireless communication technology has a long transmission distance and a slow transmission speed, the simplified data can greatly alleviate the data transmission pressure.

[0075] Specifically, the way to process the image is to extract the image features in the insect attracting points, which are called insect swarm features. The simplest feature is the data feature generated based on the image histogram.

[0076] Step S300: the master control terminal performs convolution identification on the insect swarm features according to a preset convolution kernel to obtain the type and quantity of the insect swarm; the insect swarm features are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method;

[0077] The master control terminal is a computer device with higher performance. The staff collects different insect attraction point images in advance, manually determines the insect swarm type and the number of insect swarms, and uses the same feature extraction method as step S200 to extract the insect swarm features in the insect attraction point images. The insect swarm features can be constructed into data units of the insect swarm type and the insect swarm number, and these data units are counted to obtain a table; the insect swarm features in the table are called convolution kernels.

[0078] When the swarm features uploaded by the camera module are obtained, the swarm features traverse the convolution kernel items and continuously calculate the similarity, so as to match the most suitable data unit, read the corresponding swarm type and swarm number, and use them as the swarm type and swarm number of the obtained image.

[0079] Step S400: generating a distribution image according to the type and number of insect swarms at each insect trapping point, and outputting the image as a monitoring result;

[0080] Each insect attraction point can identify an insect swarm type and number. By counting these swarm types and numbers and arranging them according to the location of the insect attraction point, a distribution image reflecting the insect swarm distribution can be obtained, which is output as a monitoring result and fed back to the staff.

[0081] The overall description of the technical solution of the present invention is as follows:

[0082] The technical solution of the present invention is generally applied to a certain indoor area, such as a storage space. The staff places insect attractant boards at various positions in the storage space, and then uses the camera in the storage space (its own supporting system) to collect images of the insect attractant boards. At this time, the microprocessor in the camera recognizes the image and extracts the insect swarm characteristics. The insect swarm characteristics are simplified data. The simplified data is transmitted to the master control end through the LoRa wireless communication technology. The master control end recognizes the insect swarm characteristics, and can determine the insect swarm characteristics and the number of insect swarms at each insect attractant board, convert them into image form for representation, and feed back to the staff as monitoring results.

[0083] Figure 2This is a first sub-flow chart of the insect visual monitoring method. The step of determining the distribution characteristics and acquisition parameters of the camera module with the insect attraction point as the target includes:

[0084] Step S101: obtaining the placement location of the insect attractant and querying the coordinates of the placement location;

[0085] Step S102: Counting the coordinates of all placement positions, and determining the installation position and working parameters of the fixed camera point based on the coordinates;

[0086] Step S103: querying the rated range of the steering camera point, clustering the coordinates according to the rated range, and determining the installation position and working parameters of the steering camera point;

[0087] The process of clustering coordinates according to the rated range is:

[0088] Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates;

[0089] Calculate the center of the same type of coordinates, and based on the center, count other coordinates within the rated range as the same type of coordinates. Execute in a loop until the distance difference between the centers of two adjacent statistical processes is less than a preset value.

[0090] In an example of the technical solution of the present invention, the calibration process of the working status of the camera module is defined. First, the placement position of the insect attractant is obtained, and the placement position is expressed in the form of coordinates. By counting the coordinates of all placement positions, the installation position of the fixed camera and its working parameters can be determined. This determination process is actually free, and only requires the fixed camera to be able to capture images of the placement position in real time.

[0091] In addition, the present application also introduces a steering camera, which is an auxiliary device. Its function is to determine whether there is a collection error in the fixed camera. It does not need to monitor a certain insect attraction point in real time. It only needs to rotate continuously and obtain images of each insect attraction point at regular intervals. Generally, one steering camera corresponds to multiple fixed cameras, and the corresponding relationship is completed based on the clustering process.

[0092] Specifically, the clustering process is: first take the placement position of the insect attractant as the center, then build a circle with a pre-set range, and position the circle with the center as the center of the circle, which can be initially clustered, and then calculate the center of the clustering result. The above process is repeated to finally obtain the most appropriate clustering result. The center of the circle after clustering is completed is used as the installation position of the steering camera.

[0093] Furthermore, the goal of setting the working parameters is to enable the camera to acquire the image of the insect attractant under the working parameters.

[0094] Figure 3 The second sub-flow diagram of the insect visual monitoring method is as follows. The steps of obtaining an insect attraction point image by a camera module, extracting insect swarm features from the insect attraction point image, and uploading the insect swarm features to the master control terminal include:

[0095] Step S201: acquiring an insect attraction point image by a camera module; when acquiring the insect attraction point image, recording the time and the coordinates of the insect attraction point as a label of the insect attraction point image;

[0096] Step S202: comparing the images of the same insect attraction point according to the labels, and determining the monitoring accuracy of each insect attraction point according to the comparison results;

[0097] Step S203: when the monitoring accuracy is less than a preset accuracy threshold, an error message is generated;

[0098] Step S204: When the monitoring accuracy is not less than a preset accuracy threshold, the insect swarm features in the insect attraction point image are extracted and uploaded to the master control terminal.

[0099] Before extracting the characteristics of the insect swarm, the technical solution of the present invention can judge the correctness of the monitoring results in combination with the monitoring results of the steering camera and the monitoring results of the fixed camera. The judgment method is to compare the two corresponding images, that is, the images of the same insect attraction point obtained by the steering camera and the fixed camera respectively. By comparing the two images, it can be judged whether there is an error. The method for judging whether there is an error is very simple. The similarity of the two images is calculated. The higher the similarity, the higher the monitoring accuracy. When the monitoring accuracy is small, it means that there may be problems in the monitoring process itself (there is a problem with a camera). When the similarity is high, enter the feature extraction step, extract the insect swarm characteristics, and upload them to the main control end.

[0100] As a preferred embodiment of the technical solution of the present invention, when the monitoring accuracy is not less than a preset accuracy threshold, the step of extracting insect swarm features in the insect attraction point image and uploading it to the master control terminal includes:

[0101] When the monitoring accuracy is not less than a preset accuracy threshold, the background is removed from the insect attraction point image according to a preset background color value;

[0102] Extract three channels of the insect attraction point image after removing the background to obtain three layers;

[0103] Traverse each layer and generate a histogram;

[0104] Calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate the conversion rule;

[0105] The histogram is equalized according to the conversion rule, and three converted histograms are obtained as the swarm characteristics.

[0106] Insect swarm features are essentially image features. First, the background of the insect trap image is easy to obtain in advance. The background generally refers to the insect trap board in the image. The color value of the insect trap board in the image is obtained and used as the background color value to locate the insect trap board in the insect trap image, and the background corresponding to the insect trap board is removed. After removing the background, the image is split into three channels. The three channels can be HSV channels or RGB channels. The purpose of splitting into three channels is to make the object of each analysis a single-value image, called a layer. After converting the layer into a histogram and expanding it to the range of [0, 255], three histograms can be obtained as insect swarm features.

[0107] It is worth mentioning that the generated histogram can be denoised, for example, the standard deviation of the histogram can be calculated and the data within the range of positive and negative 3 standard deviations can be eliminated; the reference for equalization is to convert the pixel mean to 127; in addition, the three histograms finally obtained can also be merged (calculate the mean of the number corresponding to each pixel value) to obtain a histogram to further simplify the data.

[0108] Figure 4 The third sub-flow diagram of the insect visual monitoring method is as follows. The step of performing convolution recognition on the insect swarm characteristics by the master control terminal according to a preset convolution kernel to obtain the type and number of insect swarms includes:

[0109] Step S301: establishing a connection channel with a preset convolution kernel table, and reading convolution kernels in sequence; the convolution kernel table includes convolution kernel items and swarm parameter items; the swarm parameters include swarm type and swarm quantity;

[0110] Step S302: Calculate the similarity between the convolution kernel and the insect swarm characteristics, select the convolution kernel with the highest similarity, and read the corresponding insect swarm parameters;

[0111] In an example of the technical solution of the present invention, the convolution kernel is read from the convolution kernel table, the similarity between the convolution kernel and the insect swarm feature is calculated, and then the convolution kernel and its insect swarm parameter item corresponding to the maximum similarity value are selected to serve as the recognition result of the insect attraction point image. This is essentially a convolution recognition model, and this application specifically defines some of the steps therein, mainly the feature extraction process and its comparison process, which greatly simplifies the training process and application process of the convolution recognition model.

[0112] The similarity calculation process is:

[0113] In the formula, S is the similarity, exp{-x} is the composite function of the exponential function, is the distance between two histograms, H 1i is the value of the i-th column in the first histogram, H 2iis the value of the i-th column in the second histogram, n is the total number of columns, is the mean of the first histogram, is the mean of the second histogram.

[0114] The above calculation process constructs an inverse relationship between similarity and distance. The distance adopts a scheme for measuring the similarity of two statistical sample sets. The smaller the value of the distance, the more similar the two histograms are.

[0115] Figure 5 The fourth sub-flow chart of the insect visual monitoring method is as follows. The steps of generating a distribution image based on the type and number of insect swarms at each insect attraction point and outputting the image as the monitoring result include:

[0116] Step S401: Count the type and number of insect swarms at each insect trapping point;

[0117] Step S402: query the shape corresponding to the insect swarm type, and determine the size according to the number of insect swarms; wherein the relationship between the shape and the insect swarm type, and the relationship between the size and the number of insect swarms are pre-set by the staff;

[0118] Step S403: Generate a distribution image based on the statistical shapes and sizes of the insect attracting points and output it as a monitoring result.

[0119] The staff pre-sets the shape of each insect swarm type and the mapping relationship between quantity and size (generally a proportional relationship, the more the number, the larger the size). After identifying the image of each insect attraction point, the insect swarm type and quantity at each insect attraction point are obtained. With the help of the set shape and mapping relationship, the insect swarm type and quantity can be converted into a shape with size. Combined with the distribution position of the insect attraction point, all shapes are counted to obtain a map, called a distribution image, which is output as the monitoring result.

[0120] Figure 6 1 is a composition structure diagram of an insect visual monitoring system. In an embodiment of the present invention, an insect visual monitoring system, the system 10 includes:

[0121] The camera module calibration module 11 is used to determine the distribution characteristics and acquisition parameters of the camera module with the insect trapping point as the target; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera;

[0122] The insect swarm feature extraction module 12 is used to obtain an insect attraction point image through a camera module, extract insect swarm features from the insect attraction point image, and upload the insect swarm features to the master control terminal;

[0123] The convolution recognition module 13 is used for the master control terminal to perform convolution recognition on the insect swarm characteristics according to a preset convolution kernel to obtain the type and quantity of the insect swarm; the insect swarm characteristics are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method;

[0124] The result output module 14 is used to generate a distribution image according to the type and quantity of insect swarms at each insect trapping point and output it as a monitoring result.

[0125] Furthermore, the camera module calibration module 11 includes:

[0126] A coordinate query unit, used to obtain the placement location of the insect attractant and query the coordinates of the placement location;

[0127] A coordinate statistics unit, used to count the coordinates of all placement positions, and determine the installation position and working parameters of the fixed camera point based on the coordinates;

[0128] A clustering unit, used to query the rated range of the steering camera point, cluster the coordinates according to the rated range, and determine the installation position and working parameters of the steering camera point;

[0129] The process of clustering coordinates according to the rated range is:

[0130] Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates;

[0131] Calculate the center of the same type of coordinates, and based on the center, count other coordinates within the rated range as the same type of coordinates. Execute in a loop until the distance difference between the centers of two adjacent statistical processes is less than a preset value.

[0132] Specifically, the insect swarm feature extraction module 12 includes:

[0133] An image acquisition unit is used to acquire an image of an insect attraction point through a camera module; when acquiring the image of the insect attraction point, the time and the coordinates of the insect attraction point are recorded as a label of the image of the insect attraction point;

[0134] An image comparison unit is used to compare images of the same insect attraction point according to the labels, and determine the monitoring accuracy of each insect attraction point according to the comparison results;

[0135] An error reporting unit, configured to generate an error reporting message when the monitoring accuracy is less than a preset accuracy threshold;

[0136] The extraction execution unit is used to extract the insect swarm features in the insect attraction point image and upload it to the master control terminal when the monitoring accuracy is not less than a preset accuracy threshold.

[0137] Wherein, the extraction execution unit includes:

[0138] A rejection subunit, used to reject the background in the insect attraction point image according to a preset background color value when the monitoring accuracy is not less than a preset accuracy threshold;

[0139] The extraction sub-unit is used to extract three channels of the insect attraction point image after removing the background, and obtain three layers;

[0140] Traverse each layer and generate a histogram;

[0141] A rule generation subunit is used to calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate a conversion rule;

[0142] The equalization subunit is used to equalize the histogram according to the conversion rule to obtain three converted histograms as the swarm characteristics.

[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A visual monitoring method for insect conditions, It is characterized in that The method comprises: Taking the insect attracting point as the target, determining the distribution characteristics and acquisition parameters of the camera module; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera; The camera module obtains the image of the insect attraction point, extracts the insect swarm features from the image of the insect attraction point, and uploads the insect swarm features to the master control terminal; The main control terminal performs convolution identification on the insect swarm features according to a preset convolution kernel to obtain the type and number of insect swarms; the insect swarm features are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method; According to the type and quantity of insect swarms at each insect trapping point, a distribution image is generated and output as the monitoring result; The step of determining the distribution characteristics and acquisition parameters of the camera module with the insect attracting point as the target includes: Obtain the placement location of the insect attractant and query the coordinates of the placement location; Counting the coordinates of all placement positions, and determining the installation position and working parameters of the fixed camera points based on the coordinates; Query the rated range of the steering camera point, cluster the coordinates according to the rated range, and determine the installation location and working parameters of the steering camera point; The process of clustering coordinates according to the rated range is: Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates; Calculate the center of the same type of coordinates, and count other coordinates within the rated range based on the center as the same type of coordinates, and execute it cyclically until the distance difference between the centers of two adjacent statistical processes is less than the preset value; The steps of obtaining an image of an insect attraction point by a camera module, extracting insect swarm features from the image of the insect attraction point, and uploading the insect swarm features to the master control terminal include: The camera module obtains an image of the insect attraction point; when obtaining the image of the insect attraction point, the time and the coordinates of the insect attraction point are recorded as a label of the image of the insect attraction point; Compare the images of the same insect trapping point according to the labels, and determine the monitoring accuracy of each insect trapping point according to the comparison results; When the monitoring accuracy is less than a preset accuracy threshold, an error message is generated; When the monitoring accuracy is not less than a preset accuracy threshold, the insect swarm features in the insect attraction point image are extracted and uploaded to the master control terminal.

2. The method for visually monitoring insect conditions according to claim 1, It is characterized in that When the monitoring accuracy is not less than a preset accuracy threshold, the step of extracting insect swarm features in the insect attraction point image and uploading the features to the master control terminal includes: When the monitoring accuracy is not less than a preset accuracy threshold, the background is removed from the insect attraction point image according to a preset background color value; Extract three channels of the insect attraction point image after removing the background to obtain three layers; Traverse each layer and generate a histogram; Calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate the conversion rule; The histogram is equalized according to the conversion rule, and three converted histograms are obtained as the swarm characteristics.

3. The insect visual monitoring method according to claim 1, It is characterized in that The step of performing convolution recognition on the insect swarm characteristics by the master control terminal according to a preset convolution kernel to obtain the type and number of the insect swarm includes: Establishing a connection channel with a preset convolution kernel table, and reading convolution kernels in sequence; the convolution kernel table includes convolution kernel items and swarm parameter items; the swarm parameters include swarm type and swarm quantity; Calculate the similarity between the convolution kernel and the swarm characteristics, select the convolution kernel with the highest similarity, and read the corresponding swarm parameters; The similarity calculation process is: In the formula, S is the similarity, exp{-x} is the composite function of the exponential function, is the distance between two histograms, H 1i is the value of the i-th column in the first histogram, H 2i is the value of the i-th column in the second histogram, n is the total number of columns, is the mean of the first histogram, is the mean of the second histogram.

4. The method for visually monitoring insect conditions according to claim 1, It is characterized in that The step of generating a distribution image according to the type and number of insect swarms at each insect trapping point and outputting the image as a monitoring result comprises: Count the types and numbers of insect swarms at each insect trapping point; Query the shape corresponding to the swarm type and determine the size according to the number of swarms; the relationship between the shape and the swarm type, and the relationship between the size and the number of swarms are all pre-set by the staff; Based on the statistical shapes and sizes of the insect traps, a distribution image is generated and output as the monitoring result.

5. A visual insect monitoring system, It is characterized in that The system comprises: The camera module calibration module is used to determine the distribution characteristics and acquisition parameters of the camera module with the insect attraction point as the target; the distribution characteristics are determined by the position information of the camera, and the acquisition parameters include the direction, wide angle and clarity of the camera; The insect swarm feature extraction module is used to obtain the insect attraction point image through the camera module, extract the insect swarm features from the insect attraction point image, and upload the insect swarm features to the master control terminal; A convolution recognition module is used for the master control terminal to perform convolution recognition on insect swarm features according to a preset convolution kernel to obtain the type and number of insect swarms; the insect swarm features are determined by the histogram of the insect attraction point image, and the convolution kernel is an image feature extracted from the image sample preset by the staff based on the same extraction method; The result output module is used to generate a distribution image based on the type and number of insect swarms at each insect trapping point and output it as a monitoring result; The camera module calibration module comprises: A coordinate query unit, used to obtain the placement location of the insect attractant and query the coordinates of the placement location; A coordinate statistics unit, used to count the coordinates of all placement positions, and determine the installation position and working parameters of the fixed camera point based on the coordinates; A clustering unit, used to query the rated range of the steering camera point, cluster the coordinates according to the rated range, and determine the installation position and working parameters of the steering camera point; The process of clustering coordinates according to the rated range is: Randomly select a coordinate, take the coordinate as the center, and count other coordinates within the rated range as similar coordinates; Calculate the center of the same type of coordinates, and count other coordinates within the rated range based on the center as the same type of coordinates, and execute it cyclically until the distance difference between the centers of two adjacent statistical processes is less than the preset value; The insect swarm feature extraction module includes: An image acquisition unit is used to acquire an image of an insect attraction point through a camera module; when acquiring the image of the insect attraction point, the time and the coordinates of the insect attraction point are recorded as a label of the image of the insect attraction point; An image comparison unit is used to compare images of the same insect attraction point according to the labels, and determine the monitoring accuracy of each insect attraction point according to the comparison results; An error reporting unit, configured to generate an error reporting message when the monitoring accuracy is less than a preset accuracy threshold; The extraction execution unit is used to extract the insect swarm features in the insect attraction point image and upload it to the master control terminal when the monitoring accuracy is not less than a preset accuracy threshold.

6. The insect visual monitoring system according to claim 5, It is characterized in that The extraction execution unit comprises: A rejection subunit, used to reject the background in the insect attraction point image according to a preset background color value when the monitoring accuracy is not less than a preset accuracy threshold; The extraction sub-unit is used to extract three channels of the insect attraction point image after removing the background, and obtain three layers; Traverse each layer and generate a histogram; A rule generation subunit is used to calculate the pixel mean of each histogram, convert the pixel mean to 127, and generate a conversion rule; The equalization subunit is used to equalize the histogram according to the conversion rule to obtain three converted histograms as the swarm characteristics.

Citation Information

Patent Citations

  • Pest monitoring system and method

    CN105900954A

  • Intelligent pest situation monitoring method and intelligent pest situation monitoring device

    CN107711762A

  • Global insect condition monitoring system and design method thereof

    CN111736555A

  • Solenopsis invicta prevention effect evaluation method

    CN114698613A

  • Storage grain pest multi-dimensional information acquisition and early warning method based on mobile terminal

    CN115100471A