Disinfection garage position monitoring method and monitoring system based on sign recognition

Abstract

The invention discloses a method and a system for monitoring disinfection garage positions based on mark recognition, and belongs to the technical field of image processing. The method is characterized by comprising the following steps: step a, arranging and marking; b, carrying out image acquisition through a camera; c, judging whether the camera belongs to an infrared mode or not; step d, executing a monitoring process based on a color pixel value in a non-infrared mode; step e, executing a contour-based monitoring process in a non-infrared mode; and f, sending the image processed by the monitoring process based on the contour into a mark detection model, comparing results, and determining whether the parking space is occupied or not. According to the monitoring method and monitoring system for the disinfection garage based on mark identification, the mode of directly identifying the disinfection vehicles is replaced by ground mark identification, when different disinfection vehicles are stored in the garage, state detection can be achieved, model adjustment according to the types of the stored disinfection vehicles is not needed, and the working efficiency is improved. The system is especially suitable for monitoring the storage location occupation condition in a hospital disinfection supply center.

Description

Technical Field

[0001] The invention discloses a disinfection garage location monitoring method and monitoring system based on mark recognition, belonging to the technical field of image processing. Background Art

[0002] With the continuous development of intelligent disinfection supply centers, the degree of automation in hospital disinfection supply centers is getting higher and higher. The normal operation of hospital disinfection supply centers through automated equipment can not only reduce labor costs, but also become an important means for hospital supply rooms to prevent infection and control costs. As an advanced logistics technology, AGV (automatic guided vehicle) can accurately and efficiently transport disinfected items from the supply room to various departments according to preset procedures and routes, and at the same time recycle used items to the disinfection supply room for processing. Therefore, it has gradually been used in hospital disinfection supply centers.

[0003] When there are a large number of AGVs in the disinfection supply center, it is necessary to set up special storage locations to facilitate the operation and dispatch of AGVs in the disinfection supply center. However, there are also storage locations that are occupied in the disinfection supply center. Therefore, accurately judging the usage status of the storage locations to facilitate AGV parking is an important guarantee for the smooth operation of the disinfection supply center.

[0004] In the prior art, the usage status of storage spaces in the disinfection supply center mainly relies on manual inspection and record-keeping. In a high-load working environment, it is difficult to quickly and accurately understand the actual situation of each disinfection garage space. When used in conjunction with AGV, it is not only inefficient but also prone to errors and omissions.

[0005] In the prior art, there are also many technical solutions for automatic monitoring of parking spaces through images. However, in the prior art, when judging whether a parking space is occupied, although the specific implementation methods vary, the overall idea is: aiming at the relatively fixed characteristics of the shape of the vehicle model and the shape of the parking space, then collect images of the storage space, and judge whether the storage space is occupied by judging the outline of the vehicle body in the storage space. In the disinfection supply center, the various types of disinfection vehicles towed by AGV have different shapes due to their different functions. Therefore, it is difficult to directly apply conventional image recognition methods to the monitoring of storage spaces in the disinfection supply center. Once the judgment is wrong, it is easy for the AGV to park abnormally, which poses certain safety hazards. Summary of the invention

[0006] The technical problem to be solved by the present invention is: to overcome the shortcomings of the prior art and provide a method of replacing direct identification of disinfection vehicles with ground sign recognition. When different disinfection vehicles are stored in the storage space, status detection can be achieved without the need to adjust the model according to the type of disinfection vehicles stored. The method and system are particularly suitable for disinfection garage space monitoring based on sign recognition in hospital disinfection supply centers.

[0007] The technical solution adopted by the present invention to solve the technical problem is: the disinfection garage location monitoring method based on sign recognition is characterized by comprising the following steps: Step a, arranging cameras corresponding to the storage locations, and arranging symbol marks and tag marks in the storage locations; Step b, collecting images through a camera, extracting the area corresponding to the storage location in the image to obtain a ROI image, and taking the area corresponding to the storage location as the target area; Step c, determining whether the camera is in infrared mode, if in infrared mode, executing the infrared mode monitoring process, if not in infrared mode, executing step d; Step d, executing a monitoring process based on color pixel values ​​in a non-infrared mode. If the monitoring process based on color pixel values ​​does not lead to a conclusion that the storage location corresponding to the target area is occupied, executing step e; Step e, executing the profile-based monitoring process in the non-infrared mode. If the profile-based monitoring process does not lead to a conclusion that the storage location corresponding to the target area is occupied, executing step f; In step f, the image processed by the contour-based monitoring process is sent to the mark detection model, and the mark detection model outputs the detection result. If the detection result output by the mark detection model is the same as the recorded symbol mark, it means that the storage location corresponding to the target area is not occupied. If the detection result output by the mark detection model is different from the recorded symbol mark, it means that the storage location corresponding to the target area is occupied.

[0008] Preferably, in step c, the infrared mode monitoring process is: determine whether there is a Tag mark in the target area. If there is a Tag mark, it means that the storage location corresponding to the target area is not occupied. If there is no Tag mark, it means that the storage location corresponding to the target area is occupied.

[0009] Preferably, in step d, the monitoring process based on color pixel values ​​is: Step d-1, preprocessing the ROI image, then taking the color corresponding to the symbol mark as the target color, and performing spatial extraction of the target color in the target area; Step d-2, determine whether the pixel value of the target color in the target area exceeds the threshold. If it does not exceed the threshold, it means that the storage location corresponding to the target area is occupied. If it does not exceed the threshold, execute step e.

[0010] Preferably, in step d-1, when performing spatial extraction of the target color, the RGB spatial image is converted into an HSV spatial image, and the number of pixels of the color corresponding to the symbol mark in the target area is extracted based on the hue value, saturation value and tone value.

[0011] Preferably, in step e, the profile-based monitoring process is: Step e-1, performing morphological processing on the ROI image, and then extracting the image's outline circumscribed rectangle; Step e-2, calculating the area of ​​the circumscribed rectangle; Step e-3, determine whether the area of ​​the circumscribed rectangle exceeds a threshold. If it does not exceed the threshold, it indicates that the storage location corresponding to the target area is occupied. If it does not exceed the threshold, execute step f.

[0012] Preferably, in step a, the symbol mark is a character representation of a single color, and the tag mark is a combined mark of two patterns of contrasting colors stacked one above the other.

[0013] Preferably, in step e-3, if the area of ​​the circumscribed rectangle exceeds 3 / 4 of the area of ​​the ROI image, it means that the area of ​​the circumscribed rectangle does not exceed the threshold.

[0014] A disinfection garage monitoring system based on sign recognition, characterized in that: it includes a camera preview module for realizing camera picture preview and current image acquisition, and in the early stage of implementation, adjusting the camera angle when installing the camera, obtaining the camera acquisition results and checking the status of the disinfection vehicle in real time; Camera information management module, used to configure camera parameters; The storage location information management module is used to configure storage location parameters, including the coordinates of the ROI image, the corresponding area, identification marks and storage location information notes; Algorithm parameter configuration module, used to configure algorithm parameters during state recognition; Monitoring parameter configuration module, used to configure monitoring parameters, including monitoring time interval, batch collection quantity, database login information and communication parameters; Log recording and viewing module, used to record logs and check errors; The storage location monitoring module is used to detect the collected real-time signs in the model to determine the status of the current storage location and provide status feedback; The sample training module is used to train the current logo through training samples. If the current logo is misidentified, the misidentified picture can be used as a sample, the correct label is input for it and retrained to obtain a logo detection model.

[0015] Compared with the prior art, the present invention has the following beneficial effects: In the disinfection garage space monitoring method and monitoring system based on sign recognition, ground sign recognition is used instead of direct identification of disinfection vehicles. When different disinfection vehicles are stored in the storage space, status detection can be achieved without adjusting the model according to the type of disinfection vehicles stored. It is particularly suitable for monitoring storage space occupancy in hospital disinfection supply centers.

[0016] In the disinfection garage monitoring method and monitoring system based on sign recognition, the sign detection model is a pre-trained model and there is no need to re-collect sample images for training according to the on-site environment.

[0017] In this disinfection garage monitoring system based on sign recognition, the hardware requirements are not high. When there are many network monitoring cameras, simultaneous detection occupies less computer resources and there is no need to purchase a dedicated visual server.

[0018] In the disinfection garage space monitoring method and monitoring system based on sign recognition, single-color symbol signs and tag signs are arranged in the storage space, and different monitoring processes are arranged according to whether there is sufficient light or not, without the need to add additional light sources or keep the lights on. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The flowchart of the disinfection garage location monitoring method based on sign recognition is shown in FIG.

[0020] Figure 2 This is a schematic diagram of the disinfection garage location monitoring method system based on sign recognition. DETAILED DESCRIPTION

[0021] Figure 1~2 The best embodiment of the present invention is shown below in conjunction with the attached Figure 1~2 The present invention is further described.

[0022] like Figure 1 As shown, a disinfection garage location monitoring method based on sign recognition (hereinafter referred to as monitoring method) includes the following steps: Step 1001, start.

[0023] Start to execute this monitoring method. Before executing this monitoring method, first arrange the camera, and the camera is preferably implemented by a network camera. The camera is installed above the storage location. If there are two adjacent storage locations, the camera is installed between the two adjacent storage locations. A single camera can detect the two storage locations. If there are no two adjacent storage locations, the camera is installed directly above the storage location to detect the status of the storage location to prevent detection abnormalities or failures caused by installation position problems. After completing the installation of the camera, configure the camera network and form a local area network with the server, and assign a network address to each camera.

[0024] After the camera is installed and configured, the positioning corner stickers are pasted on the four corners of the storage location, and the identification mark is pasted in the center. In this monitoring method, the identification marks set in the middle of the storage location include two types: a single-color symbol mark and a tag mark. The single-color symbol mark preferably uses red numbers, and the tag mark is an image mark containing two contrasting colors, such as a white background and a black pattern set on a white background. Among them, the single-color symbol mark is used in an environment with sufficient lighting conditions, while the tag code mark recognition is used in an environment without lighting conditions at night.

[0025] Single-color symbol signs and tag signs are pasted on the floor of the warehouse in the form of stickers, and frosted stickers are preferably used to prevent recognition failures caused by reflections. Infrared mode can be used for sign recognition in the absence of light, and the camera working mode can be set to a combination of daytime and nighttime modes for sign recognition. When using single-color symbol sign recognition, multiple consecutive digits can be used to improve stability.

[0026] Step 1002, collecting storage location images and obtaining ROI images; The image of the storage location is obtained through the camera. Among them, a single image contains multiple storage locations, and the storage location position needs to be calibrated so that the corresponding storage area position can be obtained through the calibration information during subsequent detection. The calibration method is: after pasting all the signs and positioning corner stickers of the storage location, remove all objects on the ground of the storage location; open the image obtained by the camera, and click the upper left corner and lower right corner of the sign through mouse interaction, select the point in the upper left corner of the circumscribed rectangle of the sign, whose coordinates are P1 (x1, y1), and select the point in the lower right corner, whose coordinates are P2 (x2, y2), and extract the ROI area based on these two coordinates. After adding all the storage location information, the system will establish a thread pool according to the number of storage locations, and collect and detect all storage location images at the same time, and repeat the collection and detection periodically with the collection interval set by the user.

[0027] Step 1003, determining whether the camera is in infrared mode; Determine whether the camera is in infrared mode. If so, execute step 1004. If not, execute step 1005.

[0028] Step 1004, determine whether there is a Tag mark in the image.

[0029] Determine whether there is a Tag mark in the image corresponding to the storage location. If there is a Tag mark, execute step 1013; if there is no Tag mark, execute step 1012.

[0030] Step 1005, ROI image preprocessing; Preprocess the ROI image. First, use bilateral filtering to filter the acquisition results without affecting the edge of the mark to reduce the impact of image noise.

[0031] Step 1006, color space extraction; First, the color space needs to be converted, and the RGB space image is converted into the HSV space image. According to the hue value, saturation value and tone value, the number of pixels corresponding to the color of the symbol mark in the target area is extracted. Taking red as an example, its hue range is [0-10], [156-180], the saturation range is [43-255], and the tone range is [46-255]. The red pixel value is extracted according to the above range.

[0032] Step 1007, whether the number of pixels is greater than a threshold; Determine whether the number of pixels of the corresponding color of the symbol mark in the target area (storage area) exceeds the preset threshold V threshold1 , if it exceeds the threshold V threshold1 , execute step 1008, exceed the threshold value V threshold1 , it means that the symbol mark in the target area is blocked, and step 1012 is executed.

[0033] Step 1008, performing morphological processing on the image; The binary image is opened or eroded to further improve the stability of the mark recognition, and then the outline circumscribed rectangle of the binary image is extracted and the area of ​​the rectangle is calculated.

[0034] Step 1009, determining whether the area of ​​the circumscribed rectangle is greater than a threshold; Determine whether the area of ​​the circumscribed rectangle is greater than a threshold value, where the threshold value is set to 3 / 4 of the ROI image area. If the area of ​​the circumscribed rectangle is greater than 3 / 4 of the ROI image area, execute step 1009; if the area of ​​the circumscribed rectangle is less than or equal to 3 / 4 of the ROI image area, execute step 1012.

[0035] Step 1010, model prediction output result; The morphologically processed image is detected, wherein the landmark detection model is pre-trained, and then the morphologically processed image is input into the landmark detection model for processing, and the landmark detection model outputs the detection result.

[0036] The training process of the landmark detection model is: Create a sample data set, collect multiple pictures of numbers 0-9 and Tag symbols, and define corresponding labels for them.

[0037] After pasting the corresponding identification mark label and collecting the picture containing the disinfection garage space, it is cropped and the ROI image is taken out as a training sample.

[0038] After pasting the identification mark to the disinfection garage space, select the corresponding camera and collect the image from the camera.

[0039] Select the ROI area. Specifically, through mouse interaction, select the point at the upper left corner of the circumscribed rectangle, whose coordinates are P1 (x1, y1), and select the point at the lower right corner, whose coordinates are P2 (x2, y2), and extract the ROI area based on these two coordinates.

[0040] The ROI image is preprocessed, specifically, filtering, color space extraction, morphological processing, and its length and width are uniformly adjusted to (15, 40).

[0041] Classify the sample images and define their corresponding labels. Input all processed sample images into the model and output the training results.

[0042] Step 1011, whether the output result is the same as the target result; Determine whether the symbol (such as a number) in the target area identified by the sign detection model is the same as the storage location number recorded in the system. If they are the same, execute step 1013; if they are different, execute step 1012.

[0043] Step 1012 indicates that the storage location corresponding to the target area is occupied.

[0044] Step 1013 indicates that the storage location corresponding to the target area is not occupied.

[0045] Through Figure 2 The monitoring system shown in the figure implements the above monitoring method. The monitoring system detects the status of the disinfection garage in real time and feeds back the results to the dispatching system when the status information changes, thereby realizing the dispatching control of the AGV. The hardware part mainly includes a camera switch, a server and a handheld PDA. The software system is installed on the server and mainly includes the following modules: The camera preview module, whose main function is to realize the camera screen preview and the acquisition of the current image, is used to adjust the camera angle when installing the camera in the early stage of implementation, obtain the camera acquisition results, and check the status of the disinfection vehicle in real time during later maintenance.

[0046] The camera information management module has the main function of configuring camera parameters, including camera IP, login password, port number, area location information, identification method, etc., where the identification method is the method of configuring logo recognition.

[0047] The storage location information management module has the main function of configuring the disinfection garage location parameters, including ROI coordinate information, corresponding area, identification mark and storage location information notes, etc. The identification mark is the comparison value of the model detection output result.

[0048] The algorithm parameter configuration module mainly configures the algorithm parameters for state recognition, including preprocessing algorithm parameters, comparison parameters, etc.

[0049] The monitoring parameter configuration module is used to configure monitoring parameters, including monitoring time interval, batch collection quantity, database login information and communication parameters.

[0050] The log recording and viewing module is used to record logs and check errors.

[0051] The storage location monitoring module is used to collect real-time sign detection models to determine the status of the current storage location and provide status feedback.

[0052] Sample training module, if the current sign is misidentified, the misidentified picture can be used as a sample, the correct label is input for it and retrained to obtain a new recognition model.

[0053] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A disinfection garage location monitoring method based on sign recognition, characterized in that: The steps include: Step a, arranging cameras corresponding to the storage locations, and arranging symbol marks and tag marks in the storage locations; Step b, collecting images through a camera, extracting the area corresponding to the storage location in the image to obtain a ROI image, and taking the area corresponding to the storage location as the target area; Step c, determining whether the camera is in infrared mode, if in infrared mode, executing the infrared mode monitoring process, if not in infrared mode, executing step d; Step d, executing a monitoring process based on color pixel values ​​in a non-infrared mode. If the monitoring process based on color pixel values ​​does not lead to a conclusion that the storage location corresponding to the target area is occupied, executing step e; Step e, executing the profile-based monitoring process in the non-infrared mode. If the profile-based monitoring process does not lead to a conclusion that the storage location corresponding to the target area is occupied, executing step f; In step f, the image processed by the contour-based monitoring process is sent to the mark detection model, and the mark detection model outputs the detection result. If the detection result output by the mark detection model is the same as the recorded symbol mark, it means that the storage location corresponding to the target area is not occupied. If the detection result output by the mark detection model is different from the recorded symbol mark, it means that the storage location corresponding to the target area is occupied.

2. The disinfection garage location monitoring method based on sign recognition according to claim 1 is characterized in that: In step c, the infrared mode monitoring process is: determine whether there is a Tag mark in the target area. If there is a Tag mark, it means that the storage location corresponding to the target area is not occupied. If there is no Tag mark, it means that the storage location corresponding to the target area is occupied.

3. The disinfection garage location monitoring method based on sign recognition according to claim 1 is characterized in that: In step d, the monitoring process based on color pixel values ​​is: Step d-1, preprocessing the ROI image, then taking the color corresponding to the symbol mark as the target color, and performing spatial extraction of the target color in the target area; Step d-2, determine whether the pixel value of the target color in the target area exceeds the threshold. If it does not exceed the threshold, it means that the storage location corresponding to the target area is occupied. If it does not exceed the threshold, execute step e.

4. The disinfection garage location monitoring method based on sign recognition according to claim 3 is characterized in that: In step d-1, when performing spatial extraction of the target color, the RGB spatial image is converted into an HSV spatial image, and the number of pixels of the color corresponding to the symbol mark in the target area is extracted according to the hue value, saturation value and tone value.

5. The disinfection garage location monitoring method based on sign recognition according to claim 1 or 3 is characterized in that: In step e, the profile-based monitoring process is: Step e-1, performing morphological processing on the ROI image, and then extracting the image's outline circumscribed rectangle; Step e-2, calculating the area of ​​the circumscribed rectangle; Step e-3, determine whether the area of ​​the circumscribed rectangle exceeds a threshold. If it does not exceed the threshold, it indicates that the storage location corresponding to the target area is occupied. If it does not exceed the threshold, execute step f.

6. The disinfection garage location monitoring method based on sign recognition according to claim 1 is characterized in that: In step a, the symbol mark is a character representation of a single color, and the tag mark is a combined mark of two patterns of contrasting colors stacked one above the other.

7. The disinfection garage location monitoring method based on sign recognition according to claim 5 is characterized in that: In step e-3, if the area of ​​the circumscribed rectangle exceeds 3 / 4 of the area of ​​the ROI image, it means that the area of ​​the circumscribed rectangle does not exceed the threshold.

8. A monitoring system for implementing the disinfection garage location monitoring method based on mark recognition according to any one of claims 1 to 7, characterized in that: It includes a camera preview module, which is used to realize the camera screen preview and the acquisition of the current image. In the early stage of implementation, the camera angle is adjusted when the camera is installed, and the camera acquisition results are obtained and the status of the disinfection vehicle is checked in real time; Camera information management module, used to configure camera parameters; The storage location information management module is used to configure storage location parameters, including the coordinates of the ROI image, the corresponding area, identification marks and storage location information notes; Algorithm parameter configuration module, used to configure algorithm parameters during state recognition; Monitoring parameter configuration module, used to configure monitoring parameters, including monitoring time interval, batch collection quantity, database login information and communication parameters; Log recording and viewing module, used to record logs and check errors; The storage location monitoring module is used to detect the collected real-time signs in the model to determine the status of the current storage location and provide status feedback; The sample training module is used to train the current logo through training samples. If the current logo is misidentified, the misidentified picture can be used as a sample, the correct label is input for it and retrained to obtain a logo detection model.

Images