[0028] Below, in conjunction with accompanying drawing and specific embodiment, the present invention is described further:
[0029] In order to make it convenient for everyone to carry out garbage classification processing corresponding to various types of garbage, and reduce the pressure of urban garbage disposal, the present invention provides a system that can automatically identify garbage according to the status of the existing sorted garbage being put into use. A prompt trash can that automatically recognizes the garbage classification that belongs to the classification and provides classification reminders. After the remote background configuration, the prompt garbage bins for the automatic identification of garbage classification correspond one-to-one with the categories of classified garbage bins that have been deployed and put into use. The built-in identification module uses image recognition technology to quickly identify garbage and make the corresponding classification of the garbage. Reminder for trash cans. The invention helps people quickly and conveniently classify garbage in daily life, reduces the workload of secondary sorting of garbage, enables people to easily distinguish the category of garbage and deal with it accordingly, realizes the reuse of garbage resources, and reduces the amount of garbage disposal , improve the quality of the living environment, and play a helpful role in reducing environmental pollution.
[0030] An image recognition-based garbage sorting reminder garbage system, including a garbage sorting reminder trash can, a remote management server and a management terminal, the garbage classification prompting trash can such as Figures 1 to 3 As shown, it includes frame, identification mark 01, camera 02, information processor 13, display screen 03, sensor 04, communication module 12, power supply module, advertising board 05, indicator light 07, distance sensing switch 09, control motor 14, Indicator light controller 15, rotating rod 16, telescopic spring 17, garbage can shell, built-in garbage can 19; power supply module includes solar panel 06, charging controller and storage battery.
[0031] The trash can shell is located at the bottom of the frame, and the built-in trash can 19 is located in the trash can shell, and the trash can shell is provided with a drop-in lid 08 corresponding to the built-in trash can 19 for observing the content of the built-in trash can 19 Whether the observation window 10 is full, and the door 11 that can take out the built-in trash can 19 after opening.
[0032] The quantity of setting up dustbin 19 is several, and the outer surface of setting up dustbin 19 of each uses different colors, is printed with each rubbish classification name, corresponding pattern. Manufactured separately corresponding to each garbage classification, with the external nesting of garbage bags of different colors corresponding to different classifications or combined with classification labels, it is used to hold the garbage corresponding to each classification. The slot cover 08 is arranged at the position of the sorting slot corresponding to each built-in garbage can 19, and acts as a cover for the sorting slot where the garbage is placed, preventing the garbage from being soaked by rainwater or overflowing due to other reasons. The mouth cover 08 is in one-to-one correspondence with several built-in trash cans 19.
[0033] The distance sensing switch 09 is located on the trash can shell, and the control motor 14 is located in the trash can shell, and is positioned above the drop cover 08 in the vertical direction. The side wall of the delivery opening cover 08 and the rotating rod 16 is fixedly connected. When the rotating rod 16 rotates, it can drive the delivery opening cover 08 to rotate; The output shaft is fixedly connected. The quantity of distance sensing switch 09 is several, and several distance sensing switches 09 correspond to one-to-one positions of several throwing port covers 08 .
[0034]The distance sensing switch 09 is used to perceive and judge the distance of the object, and send the first distance information to the control motor 14, so that the control motor 14 is started; Cover 08 is opened; Putting mouth cover 08 covers the sorting opening on the trash can shell, and the sorting opening is provided with a guide plate 18, and the guiding plate 18 is provided with a trash can 19 from the sorting opening to the inside and is inclined downward to form a slide. Way, the rubbish that is put into the built-in trash can 19 tops can slide into the built-in trash can 19, prevents that garbage from falling between the built-in trash can 19 outer frame and the built-in trash can 19, which is not convenient for sanitation workers to clean.
[0035] When the user prepares to put the garbage into the corresponding sorting port, the distance sensor switch of the port senses that there is an object approaching, and starts the control motor, which drives the rotating rod to rotate, and wraps the port cover on the rotating rod to open the port , it is convenient for the user to put the garbage into the corresponding built-in trash can from the sorting garbage slot; after the garbage is put in, the control motor is delayed for 3 seconds to reverse, and the pulling force of the telescopic spring 17 is used to cover the lid of the slot again on the sorting garbage slot.
[0036] The side of the opening cover 08 away from the rotating rod 16 is connected to the shell of the trash can through the telescopic spring 17; after the garbage is placed, the control motor 14 is set to delay for a certain period of time and then reverses, and the opening cover 08 is covered again by the pulling force of the telescopic spring 17 On the sorting port. Telescopic spring 17 is positioned at the both sides of sorting opening.
[0037] The quantity of indicating lamp 07 is several; Indicator lamp 07 is arranged on the top of putting in mouth cover 08, and each indicating lamp 07 corresponds to a putting in opening cover 08, and is arranged one by one with the rubbish classification that needs prompting, and a plurality of indicating lights 07 are all connected with The indicator light controller 15 is electrically connected.
[0038] The display screen 03 is arranged on one side of the frame, the advertising board 05 is arranged on the other side of the frame; the solar panel 06 is arranged on the top of the frame. The identification mark 01 is set on the frame. In this embodiment, the identification mark 01 is a two-dimensional code. Each prompt screen is equipped with a unique code, and the corresponding use code is arranged on the frame for easy scanning. This code is used to store the garbage bins in the remote management server corresponding to the deployment location when deploying the garbage bins, and is used to manage the advertisement delivery of the garbage bins at each delivery installation point according to the region.
[0039] like Figure 4 As shown, the camera 02 , the display screen 03 and the sensor 04 are all connected to the information processor 13 ; the information processor 13 is connected to the remote management server through the communication module 12 .
[0040] Camera 02 is located at one side of frame, and is positioned at the below of display screen 03; Camera 02 is used for obtaining picture information or video information in its shooting range; In the present embodiment, camera 02 is used for obtaining rubbish picture, and rubbish picture sent to the information processor 13. The sensor 04 is used to perceive and judge the distance of people, and send identification request information to the information processor 13 according to the distance information of people; the sensor 04 can be an infrared sensor 04 or a microwave sensor 04, and the sensor 04 The embedded device is on one side of the frame and close to the position of the camera 02. The sensor 04 can also be a button, and when the user presses the button, the identification request information is sent to the information processor 13 .
[0041] The information processor 13 is used to control the working state of the camera 02 according to the identification request information; for example, when the identification request information is less than the first preset distance, the camera 02 is controlled to shoot and obtain garbage pictures or videos, and the camera 02 sends the garbage pictures to the information processing Device 13.
[0042] The information processor 13 is also used to control the working state of the display screen 03 according to the identification request information; for example, when the identification request information is less than the second preset distance, control the display screen 03 to work and display preset display information.
[0043] The information processor 13 is connected to the Internet through the communication module 12, and connects and exchanges information with the remote management server through the Internet; the manager connects with the remote management server through the management terminal. The manager can send the preset display information to the remote management server through the management terminal, so that the remote management server sends the preset display information to the information processor 13, and the information processor 13 sends the preset display information to the display screen 03 for display. The preset display information may be content such as public service advertisement information. The communication module 12 can be a WIFI module or a GPRS module.
[0044] The identification module of information processor 13 is used for judging and generating classification result information according to rubbish pictures, and the classification result information is sent to display screen 03 for displaying, and according to classification result information, sends lighting signal to indicator light controller 15, so that The indicator light controller 15 controls the corresponding indicator light 07 to turn on the light; specifically, the garbage picture is preprocessed, feature extracted, and then compared with the standard template in its pre-stored standard template library to determine the category of the garbage picture, If the garbage to be identified consists of a plurality of different classification parts, each different classification area is marked respectively, the classification of each part is judged, and each area is marked correspondingly, and the classification result information is sent to the classification of the information processor 13. A prompting module; the classification result information includes area marking information and classification judgment results. The information processor 13 transmits the unidentifiable item pictures to the remote management server for backup storage through the communication module 12, and the remote management server regularly learns and trains such pictures and then updates them to the general category template library.
[0045] The classification prompt module is used to send the classification result information to the display screen 03 for display; as preferably, the garbage classification prompt trash can also includes a loudspeaker; the classification prompt module is also used to send the classification result information to the loudspeaker for voice prompts; the classification result information Displays and voice prompts are preset for each category. For example, recyclables use blue color blocks, text and marks, and are accompanied by recyclables icons and corresponding voice prompts; kitchen waste uses green color blocks and text marks with food waste icons and corresponding voice prompts; harmful waste uses red Blocks and word marks, with harmful garbage icons and corresponding voice prompts; other garbage uses gray color blocks and word marks, with other garbage icons and corresponding voice prompts. In this way, the marking is unified with the current mainstream garbage classification and marking, which is convenient for users to identify the corresponding classification and reduces the cost of learning and memory.
[0046] The remote management server is used to manage and configure garbage sorting and automatic identification prompt trash bins installed in various places, and configure classification management and advertisement delivery management for the prompt trash bins. When the equipment is installed, obtain the corresponding sorted trash can pictures, equipment QR codes and installation location information uploaded by the management personnel through the installation garbage sorting automatic identification prompt trash can, and store the above equipment installation information one by one. The management terminal manages the prompt trash can. When installing the prompt trash can, it sends the photo of the trash can set at the installation site, the QR code on the prompt screen and the detailed information of the installation site to the remote management server.
[0047] In this embodiment, the management terminal is divided into a configuration management terminal and a display management terminal; managers send device installation information to the remote management server through the configuration management terminal; and send/configure preset display information to the remote management server through the display management terminal.
[0048] When configuring the standard template library, the remote management server recognizes the corresponding classification of the trash can according to the existing garbage classification photos of the set sorting trash cans, that is, the corresponding classification of the built-in trash can 19, and then configures each garbage classification reminder trash can The classifications in the standard template library in the information processor 13, that is, the classifications corresponding to the reminders of the garbage bins for each garbage classification. If there are only two categories of recyclable and non-recyclable garbage bins installed, the automatic configuration will only provide classification reminders based on these two categories; Classes are set, and correspondingly, the categories that can be classified when prompted are also finer and more. Simultaneously, the remote management server is connected with the garbage sorting automatic recognition reminder garbage can through the network, regularly updates the trained standard template to the standard template library in the information processor 13, and obtains the unrecognizable garbage captured by the camera 02 of the prompt screen Item pictures, so as to further upgrade the general template library of the remote management server. During advertisement delivery management, advertisement delivery can be carried out according to the serial number of the device and the corresponding installation area.
[0049] It should be noted that the general category template library in the remote management server has classification catalogs of all garbage categories involved in this system, including the major categories of garbage classification, and various subdivided subcategories under the major categories; the information processor The catalogs in the standard template library also include the classification catalogs of all garbage categories involved, but the remote management server configures the garbage classification corresponding to the built-in trash can according to the characteristics of the device installation area, and the corresponding identification categories. There may be only Several categories of categories are called, or contain some subcategories, for identification. However, there may be new standard templates that are regularly updated in the general class template library, which can be updated to the standard template library in the information processor.
[0050] The display screen 03 is a touch screen, and the touch screen is also used to send an operation signal to the information processor 13 according to the user's operation, and the information processor 13 controls the working state of the camera 02 according to the operation signal. For example, after the user has identified a piece of rubbish, the user needs to perform a second identification. A control signal for re-identification can be sent to the information processor 13 through the touch screen, and the information processor 13 controls the camera 02 to take another shot.
[0051] The solar panel 06 is connected to the storage battery through the charging controller; the power supply module provides power for the camera 02, the display screen 03, the sensor 04 and the information processor 13. The solar panel 06 is arranged on the top of the frame and is fixedly arranged in a triangle.
[0052] The advertising board 05 is arranged on the other side of the frame, and can be used to display regularly updated outdoor posting print advertisements, and can be used to publicize the benefits of garbage classification.
[0053] The remote management server establishes a garbage classification directory, and according to the directory, obtains a large number of images of garbage items in the corresponding directory through a search engine for preprocessing, feature extraction, and recognition training, thereby establishing a general category template library. For example: Divide garbage into four categories: recyclables, kitchen waste, hazardous waste and other waste. Obtain the corresponding pictures of the items involved in the classification of recyclables, such as cartons, beverage plastic bottles, beer glass bottles, cans, etc., through search engines, and perform preprocessing, feature extraction, and recognition training on the pictures of these items In this way, the corresponding general category model library for recyclables identification is established; correspondingly, the corresponding items included in each category of kitchen waste, hazardous waste and other garbage are established in the same way as a general category template library.
[0054] When the installer installs the trash can for garbage sorting, he opens the management APP on the mobile phone, starts the equipment installation management system, scans the QR code on the equipment, submits the geographical location information of the installation point, and transmits the above equipment installation information to the remote management server, and complete device installation and activation. The remote management server receives the device installation information transmitted by the management terminal APP, performs corresponding configuration of the information processor 13 according to the device installation information, and records the corresponding installation location information of the storage device corresponding to the device code, and sends preset display information to the installed The information processor 13 of prompt screen.
[0055] When the user is ready to classify and identify discarded garbage items, the information processor 13 receives the recognition request information that the sensor 04 senses that someone has entered the set sensing distance range, and the display screen of the prompt trash can stops. The advertisement is played and the identification module of garbage classification is activated to enter the identification process.
[0056] The display screen 03 prompts the user to bring the garbage item to be identified close to the camera 02 so that the camera 02 can clearly obtain the image of the garbage item through voice and picture; Waiting for the user to adjust the position of the garbage item and the distance from the camera 02, after the camera 02 captures a clear image of the garbage item to be identified, the identification module identifies it;
[0057]The identification module obtains the pictures of the garbage items to be identified, and compares the images with the classification data in the garbage classification model library pre-stored in the system after preprocessing and feature extraction to determine the classification of the garbage to be identified. If the garbage to be identified consists of multiple parts and belongs to different classifications, after corresponding to each area mark, the classification of each part is judged respectively, and at the same time, the classification result information is sent to the classification prompt module.
[0058] The classification reminder module marks the classification result information transmitted by the recognition module one by one on the garbage pictures to be identified according to the reminder rules, and displays them on the display screen 03. It can also prompt the user to press the The marked category will sort the garbage into the sorting trash bin next to it. For example, "The part in the blue area is recyclable garbage, and the part in the green area is kitchen waste. Please classify according to the prompts. Thank you for your support for environmental protection." A complete garbage identification and classification prompting process is completed above. At the same time, there is a re-identification button at the bottom right of the screen, and the user can choose to re-identify to start identifying other garbage items that need to be classified. If the user leaves without performing any operation, the sensor 04 senses that the user leaves, and after feeding the information back to the information processor 13, the information processor 13 switches to the advertising module to control the display screen 03 to play advertising content.
[0059] Specifically, such as Figure 5 As shown, the garbage identification process that the information processor 13 is used to judge and generate the classification result information according to the garbage picture includes the following specific steps:
[0060] Step 101, obtaining garbage pictures to be identified;
[0061] Step 102, preprocessing the garbage picture to be identified to obtain the picture to be segmented;
[0062] Specifically, the preprocessing includes the following sub-steps: decoding the garbage picture, denoising and removing interference points to obtain the picture to be divided.
[0063] Filter the junk image to remove noise with pixel heights of 1 and 2, so that subsequent processing can be performed better. The process of filtering and removing interference points is a prior art, and will not be repeated here.
[0064] Step 103, performing segmentation processing on the image to be segmented to obtain the segmented image;
[0065] The segmentation process specifically includes the following sub-steps:
[0066] Step 31, segment the image to be segmented by the seed filling algorithm to obtain several sub-images;
[0067] Some of the sub-pictures contain only one kind of item in it, that is, the split picture; but some sub-pictures may contain more than one kind of item in it, that is, the items are glued, and these sub-pictures need to be further divided to get the final split image.
[0068] Step 32: Obtain feature information of the sub-picture, and determine whether the feature information is within a preset threshold range, if yes, the sub-picture is a segmented picture, otherwise, go to step 33.
[0069] The feature information includes the total points of the item in the sub-picture, the aspect ratio of the item, the shape of the item, the transparency of the item, the color of the item, the texture of the item and the pattern of the item, etc. The total point of the item is the number of pixels of the item in the sub-picture, and the aspect ratio is The aspect ratio of the item's area. Specifically, different feature information judgments are made according to the different picture features of each type of garbage.
[0070] For example: each type of item has some aspect ratio values of different sizes, but they all float within a range, and the corresponding threshold is established according to the floating ratio of the aspect ratio of this type of item, if it is not within the preset threshold range If it is inside, it means that a certain part of the object is blocked, and the length or width is not enough, so it deviates from the preset threshold.
[0071] The preset threshold is obtained by statistical analysis of the characteristic information of each type of garbage in the article.
[0072] It should be noted that in order to obtain different feature information of sub-pictures, the preprocessing of the garbage pictures to be identified is different. For example, in order to obtain feature information of object shape, the preprocessing process also includes edge detection; The acquisition and preprocessing process can be known from the prior art, and will not be repeated here.
[0073] Step 33: Obtain the vertical projection of the sub-picture, and obtain the position of the sticking items by finding valley points in the vertical projection, and further divide the sub-picture into split pictures based on this.
[0074] Step 104: normalize the size of the segmented image to a preset size by centroid alignment and linear interpolation enlargement, in order to prevent the problems of position offset, different size and unstable rotation of the items in the image.
[0075] Step 105, obtain the feature vector of the segmented picture; specifically: divide the segmented picture into N*M sub-segmented pictures, obtain the RGB values of all pixels in each sub-segmented picture, and obtain N*M dimensional feature vectors, N and M are preset numbers, and each feature vector includes RBG values of all pixels in the corresponding sub-segmented picture.
[0076] Step 106: Compare the feature vectors of the segmented picture with the standard templates in the standard template library one by one, judge the category of the segmented picture according to the feature vectors and standard templates, and generate classification result information.
[0077] There are standard templates for various types of garbage in the standard template library. Each type of garbage has hundreds of standard templates. After preprocessing, segmentation processing and feature extraction, the feature vectors of the classified garbage items in the training set are stored in files. middle. During training, it is necessary to indicate the correct value of each classified garbage item. The standard template of each kind of garbage is the standard feature vector of a single item. When building the standard template library, the pictures of the glued garbage items will not be added to the standard template library.
[0078] First, the information processing server establishes a garbage classification directory, and according to the directory, obtains a large number of garbage images in the corresponding directory through a search engine for preprocessing, segmentation, normalization, and feature vector acquisition, thereby establishing a standard template library for garbage classification and identification. For example: Divide garbage into 4 categories, namely: recyclables, kitchen waste, hazardous waste and other waste.
[0079] Obtain the corresponding pictures of the items involved in the classification of recyclables, such as cartons, beverage plastic bottles, beer glass bottles, cans, etc., through search engines, and preprocess, segment, and normalize the pictures of these items and obtain the feature vector to establish the corresponding standard template library for recyclables identification;
[0080] Correspondingly, a standard template library for identification is established in the same way for items included in each category of kitchen waste, hazardous waste, and other waste.
[0081] After obtaining the image of garbage to be identified, preprocessing, segmentation, and feature extraction are performed on the image and compared with the classification data in the standard template library for garbage classification pre-stored in the system to determine the category of garbage to be identified. If the garbage to be identified consists of multiple parts and belongs to different categories, after corresponding to each area mark, the category to which each part belongs is judged respectively, and at the same time, the above-mentioned area mark information and classification judgment results are transmitted to the garbage classification prompt module.
[0082] The classification reminder module marks the garbage sent by the garbage identification module corresponding to each area mark and classification result information, marks and displays them to the user one by one on the garbage pictures to be identified according to the prompt rules, and prompts the user to press The marked categories are sorted and disposed of as garbage. Combined with the voice prompt, "The part in the blue area is recyclable garbage, and the part in the green area is kitchen waste. Please sort and dispose according to the prompt marks. Thank you for your support for environmental protection."
[0083] Further, in step 32, after preliminarily judging that the sub-picture is a segmented picture, in order to prevent misjudgment, step 106 also includes judging whether the segmented picture contains sticky objects according to the feature vector and the standard template, and if so, returns to step 33, and the Segment the image for segmentation, otherwise judge the category of the segmented image based on the feature vector and the standard template.
[0084] Those skilled in the art can make various other corresponding changes and deformations according to the above-described technical solutions and concepts, and all these changes and deformations should fall within the protection scope of the claims of the present invention.
