CN110428019B - Intelligent garbage classification method and modularized intelligent garbage classification processing system - Google Patents

Abstract

The invention relates to an intelligent garbage classification method and a modularized intelligent garbage classification processing system, wherein the classification method comprises the following steps: (1) obtaining garbage data to be classified. (2) And preprocessing the obtained garbage data to be classified. And obtaining junk data which is favorable for server identification. (3) And sending the junk data which is favorable for the server to identify to a server (4) and sending the information for completing the judgment of the junk category to a user, and prompting the user about the junk category to be classified. The processing system of the invention consists of a cloud computing server, a CPU processor for calculating data, a plurality of garbage cans, a display screen capable of inputting instructions and a sensor for generating image, sound and electromagnetic data. The modularized garbage classification system is designed, the number and the types of garbage storage modules can be defined by a user to provide higher flexibility, the cost is low, and the user experience can be improved well.

Description

Intelligent garbage classification method and modularized intelligent garbage classification processing system

Technical Field

The invention relates to an intelligent garbage classification method and an intelligent garbage classification processing system for completing garbage classification by using the method.

Background

The garbage classification is an important ring of garbage recycling, the accurate classification can greatly save the cost of garbage recycling treatment, improve the garbage treatment efficiency and the garbage utilization rate, and the state is greatly pushing the garbage classification at present, however, the resistance for pushing the garbage classification is huge, and the reason is that the knowledge popularity of the garbage classification in the masses is insufficient and the awareness of the garbage classification is lower; secondly, the garbage classification rules are not completely the same due to different garbage treatment capacities of different cities, so that the popularization of garbage classification knowledge in the public and the cultivation of garbage classification consciousness are more difficult; thirdly, if the garbage classification rule is changed due to the improvement of garbage disposal capability, new knowledge is popularized again in the people and new habits are developed, which clearly means huge cost and time expenditure.

In order to solve the above problems, patent CN 109969639A and CN 109516037A propose schemes for identifying garbage by machine vision and helping users to classify, but the types of garbage that can be classified are limited by solely acquiring image information of garbage by machine vision.

The problems and the disadvantages of the prior art are that firstly, a clear garbage classification system structure is not defined, secondly, the design of the garbage classification system is not modularized, the requirements of the garbage classification occasions on the number, the form and the kind of the garbage storage modules and the data acquisition modules are different, and the non-modularized design cannot adapt to changeable application scenes; thirdly, the image acquired by the camera is not fused with the information acquired by other sensors by using an information fusion technology, and is identified by using a relevant technology (such as a deep neural network identification technology) related to machine vision, so that the acquired and utilized information is limited, and the garbage types and the upper limit of classification performance which can be classified by the system are limited; fourth, all schemes do not describe a protection method of user privacy, which is very important in intelligent classification, especially in a household intelligent garbage classification system, and is a necessary means for ensuring user information security.

In addition to the above drawbacks, the conventional trash recognition device has the following drawbacks:

the first, garbage identification device is relatively costly.

Because the key CPU and the integrated CPU are needed by the garbage identification device, and various information of image, sound and metal detection obtained by the sensor of the garbage identification device are identified, a relatively high CPU calculation chip is required to complete operation, and therefore, a CPU calculation chip and a calculation circuit main board with higher cost are required, and the cost of the garbage identification device is very high. For example, to achieve a CPU calculation chip and a calculation circuit main board for classifying thousands of kinds of garbage which may be generated in daily life normally, at the current production cost, at least 1000-2000 Yuan Renzhen, and the garbage recognition device equipped with the CPU calculation chip and the calculation circuit main board has an overall cost of at least 3000 Yuan Renzhen after other structures are included. Therefore, the price is a small burden for most families in China, and the device is difficult to popularize to most families in China.

Secondly, although the user has already classified the garbage at his home, when he packs the obtained classification and mentions the garbage transfer station (for example, when the user mentions the garbage which has already been classified at home to the garbage classification transfer station of the district), the garbage classification transfer station of the district needs to classify the garbage provided by the user again, and then the user is allowed to put the garbage provided by the user into the corresponding garbage can of the district transfer station, during which the user needs to wait for the district transfer station to complete classification analysis of the garbage, and the user can normally put the garbage.

For example, when a user lifts two garbage bags with cans and paper scraps, which are classified, to a garbage transfer station in a district, the garbage transfer station in the district needs to identify the two garbage bags again, whether the two garbage bags are identified manually or automatically by a computer system, and each garbage bag needs to take at least 50 seconds. Therefore, a garbage is classified and identified twice, not only is the resource waste and the energy waste caused, but also the energy and the time of the user can be occupied when the garbage is classified again, and bad experience is caused for the garbage loss of the user.

Therefore, there is a need for a more rationally designed garbage classification method, and an intelligent processing system that can accomplish the garbage classification method.

Disclosure of Invention

In order to solve the problems, the invention provides an intelligent garbage classification method and a modularized intelligent garbage classification processing system. The intelligent garbage classification method comprises the following steps:

(1) Obtaining garbage data to be classified;

(2) Preprocessing the obtained garbage data to be classified; obtaining junk data which is favorable for server identification;

(3) The junk data which is favorable for the identification of the server is sent to the server, intelligent identification is carried out through the server, and the classification judgment of the junk to be classified is completed; the method comprises the steps of sending the junk data which is favorable for server identification to a server, specifically, sending the junk data which is favorable for server identification to a server with highest efficiency index according to the processing efficiency index of each server;

(4) And the server sends the information for completing the judgment of the class of the garbage to the user and prompts the user of the class of the garbage to be classified.

The above-mentioned intelligent garbage classification method further comprises that the garbage data to be classified comprises one or a combination of images, sounds, metal detector detection data, infrared images, microwave radar, laser radar detection data and X-ray imaging data of the garbage to be classified; where an image refers to one of a picture and a video.

The intelligent garbage classification method is further described as follows, the preprocessing refers to preprocessing an image of garbage to be classified, and the preprocessing method comprises the following steps:

(1) Guiding a user to create conditions favorable for acquiring junk data by an acousto-optic means; acquiring a plurality of images of garbage to be classified, deleting the images of garbage to be classified which are unfavorable for intelligent recognition of the server, and re-acquiring a plurality of images when the acquired images of garbage to be classified do not meet the requirement for intelligent recognition of the server, wherein the standards for intelligent recognition of the server are as follows: the expected success rate of the identification of the image by the server is more than 60%; obtaining a final selected image;

(3) Filtering a plurality of groups of garbage data collected to the same garbage, filtering out wild values in a plurality of times of measurement, and calculating an unbiased estimated value of the measurement based on a statistical model of the data; the accuracy of this unbiased estimate should be higher than the accuracy of a single measurement; and obtaining high-precision garbage data, and finishing pretreatment.

The intelligent garbage classification method further comprises the steps of separating garbage images from the background after obtaining the final selected images, wherein the garbage classification method comprises the following steps:

(1) Acquiring garbage background data, including the background of the garbage image, and the light intensity in the background; when the background is static, directly storing the background; when the background is dynamically changed, modeling the dynamic background through a statistical model, and storing the form and parameters of the statistical model as the background;

(2) Performing target detection on the selected image, and separating the garbage from the background to obtain an image only containing the garbage;

the above-mentioned intelligent garbage classification method further describes that the screened garbage data to be classified is sent to a server, specifically: and determining the type of the transferred data according to the processing capacity of the server, transferring only static data consisting of the single measurement results of the photo and the metal detector when the processing capacity of the server is weak, and transferring the time sequence of the static data when the processing capacity of the server is strong.

The intelligent garbage classification method is further described in the following, the server adopts an information fusion algorithm based on the Bayesian theory to comprehensively process garbage data to be classified.

The intelligent garbage classification method further includes the steps of prompting the user about the type of garbage to be classified, and including the following steps:

(1) And sending the information of completing the classification judgment of the garbage to garbage classification terminal equipment used by the user, wherein the terminal equipment guides and helps the user to throw the garbage into the garbage bin of the class to which the garbage belongs. The terminal equipment marks the packaging bags of various garbage, records at least the belonging classification of the garbage, and the label also comprises a unique identification code, wherein the unique identification code is one of two-dimensional codes, bar codes, numbers, graphic codes and characters and combinations;

(2) Simultaneously giving a unique identification code to the garbage transfer equipment;

(3) When a user takes the packaging bag filled with the garbage to the garbage transfer equipment, the garbage transfer equipment identifies the garbage classification in the packaging bag of the garbage through the unique identification code in the label, and guides the user to throw the garbage in the packaging bag of the garbage into the corresponding garbage can.

The invention provides a modularized intelligent garbage classification processing system which consists of a server module, a data management module, a garbage storage module, a data recording module and a man-machine interaction module;

The server module is a cloud computing server; the data management module comprises a CPU processor for calculating data; the garbage storage module is a plurality of garbage cans; the man-machine interaction module is a display screen capable of inputting instructions; the data recording module is a sensor for acquiring images, sounds and electromagnetic data of the garbage;

the data management module is connected with the server; the garbage storage module, the data recording module and the man-machine interaction module are respectively connected with the data management module.

The modular intelligent garbage classification system described above, further described,

the server module is also provided with a first communication subsystem, and sends the information for completing the classification judgment of the garbage to the data management module;

the data management module is used for completing the processing, distribution and system control of data; executing a query and control instruction of the server according to the authorization of the user; generating a storage instruction according to the received information for completing the classification judgment of the garbage, and sending the storage instruction to a corresponding garbage storage module;

the data management module is provided with a second communication subsystem, and the second communication subsystem receives the garbage data to be classified, which is acquired by the data recording module, and sends the garbage data to the server;

The garbage storage module is used for completing garbage storage and comprises a garbage can; the garbage storage module further comprises a third communication subsystem for receiving control instructions and sending garbage can states;

the data recording module transmits the garbage data to be classified to the data management module; the sensor comprises a camera, a metal detector, a distance sensor, an illumination system for supplementing light for the camera, a radar containing microwave, a laser radar and an X-ray imager;

the man-machine interaction module displays the content provided by the data management module to the user, wherein the content comprises the statistical data of garbage throwing and the residual capacity of the garbage storage module; the man-machine interaction module receives data and instructions input by a user, and the data and instructions comprise definition of garbage types stored in the garbage storage module and system function setting; the man-machine interaction module provides data and instructions input by a user to the data management module.

The above-mentioned modularized intelligent garbage classification processing system further illustrates that the communication mode between the data management module and the server is one or a combination of a WIFI, a 4G/5G network and a wired network, and the connection mode between the data management module and the garbage storage module is one or a combination of a Bluetooth, a WIFI and a wired network; the communication is an encrypted communication; the data processing module only transmits data that can be transmitted by the user authorization to the server.

The data management module, the data recording module and the user interaction module are independent in hardware and integrated into one of single hardware devices.

The beneficial effects of the invention are as follows:

1. the invention designs a modularized garbage classification system, the number and the types of garbage storage modules can be defined by a user to provide higher flexibility, and the data management module, the data recording module and the man-machine interaction module can be realized in a separated way on hardware and can be integrated with a mobile phone and wearable equipment, so that convenience is provided for the user greatly; the server completes multi-sensor fusion identification, so that the high performance of the classifier is ensured, and the low cost of system adjustment is maintained when the classification rule is changed; the 5G is integrated into the system, and the garbage identification waiting time is shortened by utilizing the characteristics of high data transmission rate and low delay; the user privacy protection mechanism is designed, so that garbage classification convenience is provided for users, and the information security of the users is guaranteed. The system data is managed through the design data management module, so that a user can authorize the data type and the data volume transmitted to the server by the system, and the user privacy can be protected while the garbage classification convenience is provided; and the safety of user information is ensured by encrypting and transmitting communication data among the modules.

2. According to the invention, only the picture preprocessing system is arranged locally, and garbage identification work requiring more calculation is handed to a cloud server for processing, so that a chip and a main board for local preprocessing can be used as a chip and a main board with low cost, the cost of the garbage identification processing device can be very low, the cost of purchasing and bearing by individuals is adapted, the device is convenient to popularize to various public application scenes requiring garbage classification, such as individuals, communities, subway stations, roadside garbage cans and the like, and the realization of smart cities is supported in the garbage recycling field.

3. The garbage bag identifying device can be used for marking the unique identifying label on the garbage bag which is already identified, and the unique identifying label is transmitted to the garbage transfer station of the community through the Internet, when a user carries the garbage bag which is already provided with the unique identifying label to the garbage transfer station of the community, the garbage transfer station of the community can directly identify the unique identifying label, garbage classification and identification work is not needed again, the user is directly guided to put in the corresponding garbage can, the garbage putting time of the user is saved, and the user experience is improved.

Drawings

Fig. 1 is a system configuration diagram of the present invention.

Description of the embodiments

The data management module is the core of the system and completes the processing, distribution and system control of the data.

And a data management module: the data management module comprises a CPU processor (computer) for calculating data, and various data conversion circuits, a data coupling circuit, a driving circuit, a power supply transformation circuit and a protection circuit are arranged on a CPU peripheral circuit.

The computer referred to herein is a single-chip microcomputer, and the CPU processor (computer) is a central processing unit of the single-chip microcomputer chip, and is mainly used for completing arithmetic or logic operation of data, and simultaneously coordinating and managing the operation of the whole single-chip microcomputer chip system.

The single chip microcomputer is also provided with a storage area for storing data and programs, wherein the data storage area is used for storing intermediate results of the data, finishing functions of data temporary storage, data buffering and the like, and the storage area of the single chip microcomputer chip is generally divided into an on-chip storage area and an off-chip storage area, wherein the on-chip storage area can be divided into a read-only memory for storing the programs and a random access memory for storing the data. The off-chip memory is an extension of the on-chip memory and may be implemented by an erasable programmable read-only memory, an electrically erasable programmable read-only memory, or the like.

When the data management module works, the CPU (computer) for calculating the data firstly performs block compression on the data to be stored before the data is stored, then stores the data in the storage space of the computer, firstly decompresses the instructions in the storage space before executing the instructions in the storage space, and then provides the decompressed instructions for an external chip of the singlechip chip for processing. The function of performing block compression on instruction data may be performed by a compiler external to the monolithic chip, and the function of performing decompression may be performed by a processor internal to the monolithic chip.

After the processor of the singlechip chip finishes executing the current instruction, the processor judges whether the next instruction to be executed is positioned in the address range associated with the current instruction; if yes, the processor continues to execute the next instruction in the address range of the current instruction; if not, the processor extracts the corresponding instruction from the storage space again, decompresses the instruction, and then executes the instruction after solving the shrinkage.

And a data recording module:

the data recording module acquires the data of the garbage to be classified through the sensor. The data recording module comprises a plurality of sensors, the types of the sensors comprise cameras, and the sensors can also comprise metal detectors, distance sensors and other types of sensors (infrared image cameras, microwave radar, laser radar detection data, X-ray imaging data and the like) according to the types of garbage to be classified and the needs of scenes, and the data recording module also comprises an illumination system to help the cameras to acquire images of the garbage to be classified more accurately; the triggering condition of the illumination subsystem is that the distance between the garbage and the camera is within a certain range, a ranging module (the specific ranging mode can be ultrasonic ranging, infrared ranging, laser ranging or radar ranging and the like) is arranged beside the camera, the distance between the garbage and the camera is measured, and when the distance is within a set range, the camera and a flash lamp are turned on to obtain an image.

The metal detector adopts an all-metal detector: all metal impurities of different materials can be detected, and the current market share of the metal detector is highest. The alarm device of the existing all-metal detector is electrically connected to the data recording module of the invention, when metal is detected, the alarm current sent out is received into the data recording module to generate an analog signal quantity, and the data processing module marks the analog signal quantity as a metal label.

The method for acquiring images and processing data by using the camera is mature in the current technology.

According to the different processing capacity and the different recording scenes, the means of data recording are two types, one type is that a data recording module continuously records a plurality of groups of data, and garbage data which is favorable for the identification of a server is selected and transmitted to the server; the other is to use technical means to guide the user to create favorable conditions for recording the garbage data which is favorable for the server to identify, for example, a distance sensor is arranged beside a camera to measure the distance of the garbage, when the distance of the garbage is positioned in a preset distance section, the user is prompted by means of optics, acoustics or combination of the optics and acoustics, the data recording conditions are met currently, the data is recorded, the recording action refers to multiple times of measurement on the same target, and multiple groups of data of the same target are recorded.

The data recording module transmits the recorded data to the data management module. The method involves data conversion, or data conversion is carried out by a digital recording module or obtained data is converted by a data management module, and the realization method is as follows through a data conversion circuit: the data conversion circuit, a device that converts an analog quantity into a digital quantity, is called an analog-to-digital converter, and is simply called an a/D. The a/D conversion process is completed through 4 steps of sampling, holding, quantizing and encoding, which is called an analog-to-digital converter, and when the a/D conversion is finished, the ADC outputs a conversion finish signal data. The CPU can read the conversion result in a variety of ways: (1) a query mode; (2) an interrupt mode; (3) DMA mode. The present invention may employ CMOS components with 8-bit a/D converters, 8-way multiplexing switches, and microprocessor compatible control logic. The successive approximation type A/D converter can be directly connected with a singlechip. The system consists of an 8-way analog switch, an address latch and decoder, an A/D converter and a tri-state output latch. The multi-way switch can gate 8 analog channels, allows 8 analog quantity to be input in a time sharing way, and is used for conversion by the common A/D converter. The tri-state output latch is used for latching the digital quantity after A/D conversion, and the converted data is taken out from the tri-state output latch.

In the practical system, various physical parameters (such as pressure, electromagnetic induction, sound and the like) are measured by a sensor and converted into electric signals, and then the electric signals are transmitted to a CPU of a computer through an A/D converter; after the processing of the CPU, the D/A converter is used for controlling various parameter quantities.

The main work of the garbage storage module is to finish garbage storage. The main structure of the garbage storage module is a garbage can, a garbage can cover plate capable of controlling opening and closing, a mechanical structure for controlling the cover plate, a sensor for judging the garbage amount in the can, a garbage compression mechanical structure, an embedded third computer for receiving control instructions, sending a garbage can state communication subsystem and a third communication subsystem for controlling the mechanical structure; the embedded third computer receives the storage instruction sent by the data management module through the third communication subsystem, controls the barrel cover and the garbage compression mechanical structure according to the instruction, and sends the state of the garbage can to the data management module through the third communication subsystem.

The garbage collection module is implemented as follows:

the man-machine interaction module is used for displaying data to a system user, wherein the data comprise, but are not limited to, statistics data of garbage throwing and residual capacity of the garbage storage module; the man-machine interaction module can also receive data and instructions input by a user, and the content of the man-machine interaction module comprises, but is not limited to, definition of garbage types stored in the garbage storage module and system function setting.

The data management module comprises a second communication subsystem and a second computer, and has the functions of providing display content for the man-machine interaction module, receiving data and instructions input by the man-machine interaction module, controlling various states of the system according to the data and the instructions, and transmitting system data which can be transmitted by the system user authorization to the server so as to ensure the privacy of the user; the method comprises the steps of receiving garbage data to be classified, which is acquired by a data recording module, and transmitting the garbage data to a server to finish identification; receiving the identification result and inquiry and control data and instructions sent by the server, generating a storage instruction according to the identification result, sending the storage instruction to the corresponding garbage storage module, and executing the inquiry and control instructions of the server according to the authorization of the system user.

The server module comprises a high-performance server and a first communication subsystem, and has the functions of identifying the garbage type according to the garbage data to be classified, which is sent by the data management module, and sending the identification result to the data management module; system status is obtained or queried from the data management module, system data is collected and analyzed, and necessary control instructions are generated and sent.

Further, the number of the garbage storage modules and the kinds of garbage stored therein may be defined by a user. The definition of waste is determined based on local requirements for waste classification, e.g. local regulations require classification of waste into kitchen waste and recyclable waste, and the user can only define waste bins into these two categories.

Further, the garbage storage module can pack and bag stored garbage, and after the garbage storage module packs, the district garbage transfer station can identify the garbage bag and throw the garbage bag into the corresponding garbage storage module.

Furthermore, the data management module, the data recording module and the user interaction module can be integrated into a single hardware device in hardware, such as a portable device such as a mobile phone or a wearable device such as a smart bracelet and augmented reality glasses. The server module is cloud computing, and the garbage storage module is a small garbage can with a third computer. The communication modes between the data management module and the server and between the data management and garbage storage modules are 5G networks, and other communication modes with transmission rate higher than 500KB/s, such as 4G, WIFI and Bluetooth, can be adopted under the condition of no 5G. The better choices are: the communication mode between the data management module and the server can select WIFI, 4G/5G network and wired network. And the connection mode between the data management and the garbage storage module can be Bluetooth, WIFI or a wired network.

The method comprises the steps that a server identifies garbage by deep learning and information fusion (including but not limited to a deep neural network technology), the garbage image is identified by the deep learning, and then the final classification of the garbage is completed by fusing an image identification result and other sensor data by the information fusion technology. Further, the communication between the modules is encrypted communication, and a public key encryption algorithm, such as an RSA encryption algorithm, is adopted, which is that the information receiver generates a pair of public key and private key, sends the public key to the information sender, encrypts the information sent by the public key, and decrypts the encrypted information by the private key.

Server identification garbage adopts deep learning and information fusion modes (including but not limited to deep neural network technology) to identify garbage images, for example:

after the server acquires the image samples, the image samples are input into an original neural network model, and the neural network model suitable for configuration parameter information is obtained through learning training of each network in the original neural network model. According to the neural network model obtained through training of a large number of high-frequency garbage image samples, the image data can be identified rapidly and efficiently when a later user collects the image data. For example, the training process is to preprocess the image sample, extract the image feature vector of the preprocessed image sample by using the DNN feature extraction network, calculate the loss value by using the forward conduction algorithm and the loss function according to the extracted image feature vector and the image label, and optimize the parameters of each layer of DNN by using the backward conduction algorithm.

Further, the data processing module only transmits data transmittable by user authorization to the server to protect user privacy. The user privacy information comprises specific types of garbage in the garbage can, the time for throwing garbage by a user, the quantity of garbage thrown by the user and the statistical data of the three contents, and further comprises photos shot by a camera, distance sensor data, mechanical sensor data and operation records of all modules. The user opens the information transmission authorization settings through the human-machine interface, and the user authorization settings can be transmitted to the information of the server.

The Computer (CPU) of the present invention comprises a computer readable storage medium and a computer program mechanism embedded therein, the computer program mechanism comprising instructions for performing the steps of any of the methods described above.

The above-described parts of the apparatus are described separately in terms of functional division into various modules. The functions of the modules or units may be implemented in the same piece or pieces of software or hardware.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The intelligent garbage classification method comprises the following steps:

firstly, obtaining garbage data to be classified; the method comprises one of image, sound, metal detector detection data, infrared image, microwave radar, laser radar detection data and X-ray imaging data of garbage to be classified and combination; where an image refers to one of a picture and a video. The junk data is obtained through the information recording module.

The invention is mainly used for acquiring image information and mainly acquired through a camera. When other information is incomplete, only the obtained image information may be sent to the data processing module of the present invention, and of course, when only other information (sound, metal detector detection data, infrared image, microwave radar, etc.) is not included, the image information is optimally selected not to be sent to the data processing module of the present invention.

Secondly, the data processing module preprocesses the obtained garbage data to be classified; the method mainly refers to preprocessing images of garbage to be classified, and comprises the following steps:

(1) The user is guided by the acousto-optic means to create conditions favorable for acquiring the garbage data (specifically, the distance between the garbage and the camera is measured by the range finder, the brightness of the camera light supplementing lamp is controlled according to the distance, when the garbage is positioned in an ideal data acquisition area, the light supplementing lamp is brightest, and the garbage can sends a 'click' sound to inform the user that the garbage position meets the data acquisition requirement and the data is being acquired). Acquiring a plurality of images of garbage to be classified, deleting the images of garbage to be classified which are unfavorable for intelligent recognition of the server, and re-acquiring a plurality of images when the acquired images of garbage to be classified do not meet the requirement for intelligent recognition of the server, wherein the standards for intelligent recognition of the server are as follows: the expected success rate of the identification of the image by the server is more than 60%; a final selected image is obtained. For example, 10 pictures are obtained through a camera, then screening is carried out on the 10 pictures, from the first picture, if the first picture meets the requirements, the rest 9 pictures are directly deleted, if the first picture does not meet the requirements, the first picture is deleted, the second picture is continuously analyzed, and at least one picture beneficial to intelligent recognition of a server is obtained through flow analysis.

In order to ensure that the identification is more accurate, the success rate of the standard which is beneficial to intelligent identification of the server is improved to 80%; the judging method comprises the following steps: the requirements of the server on parameter values of some items of smoothness, texture, gray scale, color, spectrum characteristics, spatial characteristics, geometric shape analysis, size, color density, spectrum and the like of the photo before recording are superior to the parameter values, namely the server is considered to be identifiable, otherwise, the server is considered to be difficult to identify.

(2) Separating the garbage image from the background from the selected image, wherein the garbage image comprises the following steps:

A. acquiring garbage background data, including the background of a garbage image, the light intensity in the background, the light intensity in the view field of a camera and the like; when the background is static, directly storing the background; when the background is dynamically changed, for example, a moving object such as branches swaying with wind exists in the background, the dynamic background is modeled through a statistical model, the form and parameters of the statistical model (such as Gaussian mixture distribution) are stored, as the background, a data recording module opens a distance sensor and a camera to sense the environment according to a certain strategy, and when the measured distance is found to be changed (meaning that the camera is possibly moved), or the background, the light color, the light intensity and other environmental factors in the view of the camera are changed, garbage background data can be obtained again.

B. Performing target detection on the selected image, and separating garbage from the background to obtain an image only containing garbage;

separating the spam images from the background can increase the spam recognition rate of the server, but not just the essential components of the spam classification system, whether or not this approach is used depends on the processing power of the head-end equipment and the requirements of the server recognition algorithm.

(3) Filtering the collected multiple groups of garbage-only data, filtering out wild values in multiple measurements, and calculating an unbiased estimated value of the measurement based on a statistical model of the data; the accuracy of this unbiased estimate should be higher than the accuracy of a single measurement; and obtaining a high-precision garbage image, finishing preprocessing, and finally obtaining garbage data which is favorable for server identification. The calculation of the unbiased estimate is based on the probability distribution of the data.

At present, more methods for image processing are mature, for example: filtering (smoothing, noise reduction), enhancement, edge sharpening, texture analysis (de-skeletons, connectivity), image segmentation (gray scale, color, spectral features, texture features, spatial features), geometry analysis (Blob analysis) (shape, size, length, area, edges, circularity location, orientation, quantity, connectivity, etc.), blob analysis (number, location, shape, and orientation of blobs in an image can be provided for machine vision applications, and topology between related blobs can also be provided applications: two-dimensional target images, high contrast images, presence/absence detection, numerical range, and rotational invariance requirements), color analysis (chromaticity, color density, spectrum, automatic white balancing). The present invention may utilize some of the above methods for processing or judging background images, garbage images (foreground images).

Thirdly, the junk data which is favorable for the identification of the server is sent to the server, intelligent identification is carried out through the server, and the classification judgment of the junk to be classified is completed; the garbage data which is favorable for the identification of the server is sent to the server, specifically, the servers which are in charge of the identification are connected with each other to form a network, each data management module calculates the efficiency index of the server in real time according to factors such as the processing capacity of each server, the task queue condition, the data transmission speed and the like, and the higher the efficiency index is, the faster the identification task speed of the terminal is processed by the server, the data management module entrusts the identification task (garbage data which is favorable for the identification of the server) to the server with the highest efficiency index, so that the optimal identification efficiency is ensured.

The data management module determines the type of data to be transferred according to the processing capacity of the identification server corresponding to the data management module, when the processing capacity of the server is weak, the data management module only transfers static data, such as data composed of static photos and the identification result of the single metal detector, and if the processing capacity of the server is strong, the data management module transfers the time sequence of the static data, such as a section of video composed of a plurality of photos.

Fourthly, the server adopts an information fusion algorithm based on a Bayesian theory to comprehensively process the junk data to be classified; the server sends the information of completing the classification judgment of the garbage to the user, and prompts the user about the classification of the garbage to be classified according to the following steps:

(1) And sending the information of completing the classification judgment of the garbage to a garbage storage module (household intelligent garbage cans used by users and household garbage classification intelligent terminals), wherein the garbage cans guide and help the users to throw the garbage into the garbage cans with the classification of the garbage (for example, a display screen on the intelligent garbage cans displays images of the garbage to be thrown by the users, names of the garbage and the garbage classification, for example, the thrown garbage is fruit peel, the display screen displays the names of the garbage peel and the classification of the garbage wet). The garbage can is marked with a label on the packaging bag of various garbage, the label at least records the classification of the garbage, the label also comprises a unique identification code, and the unique identification code comprises one of a two-dimensional code, a bar code, a number, a graphic code and a character and a combination.

(2) And simultaneously, the unique identification code is given to garbage transfer equipment, such as a district garbage can (a garbage transfer equipment, and meanwhile, the garbage transfer equipment also has an intelligent garbage classification function).

(3) When a user takes the packaging bag filled with the garbage to the garbage can of the district, the garbage can of the district recognizes the garbage classification in the packaging bag of the garbage through the unique identification code in the label, and guides the user to throw the garbage in the packaging bag of the garbage into the corresponding garbage can.

Specifically, the method comprises the following steps: after the user carries the garbage bag with the unique identification tag to the community garbage can, the community garbage can directly identify the unique identification tag without carrying out garbage classification identification work again, the user is directly guided to put in the corresponding garbage can (for example, the user can directly control to open the cover of the corresponding garbage can), the garbage putting time of the user is saved, and the user experience is improved.

The experience method comprises the following steps: the server transmits the unique identification two-dimensional codes to the garbage transfer station and the garbage storage module respectively, and the garbage storage module packages garbage and then prints the unique identification two-dimensional codes on garbage bags. When the user lifts the packed garbage bags to the garbage transfer station, the garbage transfer station recognizes the unique two-dimensional code on the garbage bags, namely, the user can consider the garbage can with the cover opened as a disposable garbage can for knowing which garbage can the garbage should be placed in, and can directly put the garbage can into the garbage can.

The garbage collection module of the present invention can be implemented as follows:

a plurality of small garbage cans are spliced together to form a garbage can, and each classified garbage can is a garbage storage module required by the invention.

The garbage can body is hinged with a can cover plate, a motor with a rotating shaft is arranged at the can opening of the garbage can body, an arm rod is arranged at the bottom edge of the can cover plate, and a rotating shaft with the motor with the rotating shaft at the can opening of the garbage can body penetrates through the arm rod arranged at the bottom of the can cover plate. The top of the bin cover plate is provided with a solar panel which is electrically connected with a storage battery arranged at the bottom of the garbage bin body and used for isolating the cavity; the storage battery is electrically connected with the control circuit board, and the control circuit board is electrically connected with various sensors (an illuminating lamp, a camera, a metal detector and an infrared distance sensor) required by the invention, a motor with a rotating shaft and a display with touch operation respectively, and a Computer (CPU) required by the invention is also arranged on the circuit board.

The garbage can bottom is provided with a partition board, an isolation cavity is formed between the partition board and a can bottom plate of the garbage can, a control circuit board and a storage battery are arranged in the isolation cavity, and a metal detector is arranged at the top of the partition board.

The top at bung board sets up solar panel, and the bottom surface middle part of bung board sets up light and camera for make at closing the bung board, the interior black paint of bucket, automatic open the light provides the light source for the camera, the camera shoots the interior rubbish state of bucket and carries out information storage and be convenient for later stage and carry out data analysis.

The garbage bin is provided with a miniaturized label printer (such as a ministerial label machine B11 produced by Wuhan ministerial science and technology Co., ltd.) which is connected with a circuit board on the garbage bin and is responsible for printing a small label paper, and the label paper is provided with an identification code unique identification label.

The inner wall below the bin mouth of the garbage bin body is provided with an infrared distance sensor, the control circuit board controls the motor with the rotating shaft to rotate anticlockwise, the rotating shaft with the motor with the rotating shaft drives the arm rod to swing anticlockwise, the bin cover plate connected with the arm rod into a whole turns anticlockwise, and the bin cover plate is controlled to be opened/closed.

Claims (10)

1. The intelligent garbage classification method is characterized by comprising the following steps of:

s1, obtaining garbage data to be classified; the garbage data to be classified comprises one or a combination of images, sound, metal detector detection data, infrared images, microwave radar, laser radar detection data and X-ray imaging data of garbage to be classified; wherein the image refers to one of a picture and a video;

S2, preprocessing the obtained garbage data to be classified; obtaining junk data which is favorable for server identification;

s3, sending the junk data which is favorable for the identification of the server to the server, and performing intelligent identification through the server to finish the classification judgment of the junk to be classified; the data management module calculates the efficiency index of the server in real time according to one or a combination of the processing capacity, the task queue condition and the data transmission speed of each server, and then sends the garbage data which is favorable for the server to the server with the highest efficiency index according to the processing efficiency index of each server;

s4, the server sends the information of completing the judgment of the class of the garbage to the user, and prompts the user of the class of the garbage to be classified; specifically, the information of completing the classification judgment of the garbage is sent to garbage classification terminal equipment used by a user;

the terminal equipment guides and helps a user to put garbage into a garbage can of the class to which the garbage belongs; the terminal equipment marks the packaging bags of various garbage, records at least the belonging classification of the garbage, and the label also comprises a unique identification code; meanwhile, giving the unique identification code to garbage transfer equipment;

And S5, when the user takes the packaging bag filled with the garbage to the garbage transfer equipment, the garbage transfer equipment recognizes the garbage classification in the packaging bag of the garbage through the unique identification code in the label, and guides the user to throw the garbage in the packaging bag of the garbage into the corresponding garbage can.

2. The intelligent garbage classification method according to claim 1, wherein the preprocessing refers to preprocessing an image of garbage to be classified, and the preprocessing method comprises the following steps:

(1) Guiding a user to create conditions favorable for acquiring junk data by an acousto-optic means; acquiring a plurality of images of garbage to be classified, deleting the images of garbage to be classified which are unfavorable for intelligent recognition of the server, and re-acquiring a plurality of images when the acquired images of garbage to be classified do not meet the requirement for intelligent recognition of the server, wherein the standards for intelligent recognition of the server are as follows: the expected recognition success rate of the server to the image is more than 80 percent, and a final selected image is obtained; the judging method comprises the following steps: the requirements of the server on parameter values of projects such as smoothness, texture, gray scale, color, spectrum characteristics, spatial characteristics, geometric shape analysis, size, color density and spectrum of the photo before recording are superior to the parameter values, namely the server is considered to be identifiable, otherwise, the server is considered to be difficult to identify;

(2) Filtering a plurality of groups of garbage data collected to the same garbage, filtering out wild values in a plurality of times of measurement, and calculating an unbiased estimated value of the measurement based on a statistical model of the data; the accuracy of this unbiased estimate should be higher than the accuracy of a single measurement; and obtaining high-precision garbage data, and finishing pretreatment.

3. The intelligent garbage classification method according to claim 2, wherein after obtaining the final selected image, the garbage image is further separated from the background, and the process is as follows:

(1) Acquiring garbage background data, including the background of the garbage image, and the light intensity in the background; when the background is static, directly storing the background; when the background is dynamically changed, modeling the dynamic background through a statistical model, and storing the form and parameters of the statistical model as the background;

(2) And performing target detection on the selected image, and separating the garbage from the background to obtain an image only containing the garbage.

4. The intelligent garbage classification method according to claim 1, wherein the step of sending garbage data that facilitates server identification to a server further comprises: and determining the type of the transferred data according to the processing capacity of the server, transferring only static data consisting of the single measurement results of the photo and the metal detector when the processing capacity of the server is weak, and transferring the time sequence of the static data when the processing capacity of the server is strong.

5. The intelligent garbage classification method according to claim 1, wherein the server adopts an information fusion algorithm based on a Bayesian theory to comprehensively process garbage data to be classified.

6. The intelligent garbage classification method according to claim 1, wherein the unique identification code is one or a combination of two-dimensional code, bar code, number, graphic code and text.

7. The modularized intelligent garbage classification processing system consists of a server module, a data management module, a garbage storage module, a data recording module and a man-machine interaction module; for running the intelligent garbage classification method of any one of claims 1-6; the method is characterized in that:

the server module is a cloud computing server; the data management module comprises a CPU processor for calculating data; the garbage storage module is a plurality of garbage cans; the man-machine interaction module is a display screen capable of inputting instructions; the data recording module is a sensor for acquiring images, sounds and electromagnetic data of the garbage;

the data management module is connected with the server; the garbage storage module, the data recording module and the man-machine interaction module are respectively connected with the data management module.

8. The modular intelligent garbage classification system of claim 7, wherein,

The server module is also provided with a first communication subsystem, and sends the information for completing the classification judgment of the garbage to the data management module;

the data management module is used for completing the processing, distribution and system control of data; executing a query and control instruction of the server according to the authorization of the user; generating a storage instruction according to the received information for completing the classification judgment of the garbage, and sending the storage instruction to a corresponding garbage storage module;

the data management module is provided with a second communication subsystem, and the second communication subsystem receives the garbage data to be classified, which is acquired by the data recording module, and sends the garbage data to the server;

the garbage storage module is used for completing garbage storage and comprises a garbage can; the garbage storage module further comprises a third communication subsystem for receiving control instructions and sending garbage can states;

the data recording module transmits the garbage data to be classified to the data management module; the sensor comprises a camera, a metal detector, a distance sensor, an illumination system for supplementing light for the camera, a radar containing microwave, a laser radar and an X-ray imager;

the man-machine interaction module displays the content provided by the data management module to the user, wherein the content comprises the statistical data of garbage throwing and the residual capacity of the garbage storage module; the man-machine interaction module receives data and instructions input by a user, and the data and instructions comprise definition of garbage types stored in the garbage storage module and system function setting; the man-machine interaction module provides data and instructions input by a user to the data management module.

9. The modular intelligent garbage classification system of claim 8, further comprising a garbage transfer station, wherein the server is connected to the garbage transfer station for the server to send information to the garbage transfer station that completes the classification of garbage;

the refuse transfer station can recognize the unique identification code.

10. The modular intelligent garbage classification processing system according to claim 7, wherein the communication mode between the data management module and the server is one or a combination of WIFI, 4G/5G network and wired network, and the connection mode between the data management module and the garbage storage module is one or a combination of bluetooth, WIFI and wired network; the communication is an encrypted communication; the data processing module only transmits data which can be transmitted by the authorization of the user to the server;

the data management module, the data recording module and the user interaction module are independent in hardware and integrated into one of single hardware devices.

Images