Method and device for identifying graphic code on surface of special container in nuclear radiation environment

Abstract

The invention discloses a method and a device for identifying a graphic code on the surface of a special container in a nuclear radiation environment. The method comprises the following steps: (1) acquiring a video stream; (2) decoding the video; selecting a video frame needing to be processed; performing target detection and identification on the selected video frame, and determining a target image; performing graying processing on the target image to obtain a grayscale image; detecting whether a graphic code exists in the grayscale image, and determining the position of the graphic code; performing local matting according to the position of the graphic code to obtain a graphic code image; removing noise of the graphic code image; performing hybrid binarization processing or OSTU binarization processing on the graphic code image after noise removal to obtain a binarized image; performing graphic code recognition on the binarized image to obtain a recognition result; and (3) storing an identification result in a local storage medium and reporting the identification result to an MES system. According to the invention, real-time accurate identification and automatic intelligent efficient identification of the graphic code on the surface of the special container in the nuclear radiation environment can be realized.

Description

Technical Field

[0001] The invention belongs to the field of nuclear technology, and in particular relates to a method and a device for recognizing graphic codes on the surface of special containers in a nuclear radiation environment. Background Art

[0002] With the rapid development of artificial intelligence technology, image recognition technology has been widely used in all walks of life. As image recognition technology gradually matures and its advantages become more and more obvious, the entire nuclear energy industry, including nuclear fuel production, nuclear power plant operation and maintenance, spent fuel reprocessing, and nuclear waste storage, has gradually begun to pay attention to this emerging technology.

[0003] For a long time, there has been a lot of manual character number recognition work in the nuclear energy industry using human eyes, telescopes or television cameras. It has long been proven that graphic code recognition technology is a more efficient, accurate and economical choice for these tasks.

[0004] Graphic code automatic recognition technology has been widely used in manufacturing, commerce and the Internet of Things industry. However, it is rarely used in the nuclear energy field. This is mainly because the coding medium of graphic codes is often smooth stainless steel or other metal surfaces. The surface of the coded object is very smooth, and different angles will produce different reflected light. At the same time, the coded object needs to be burned. After burning, the contrast of the graphic code will decrease, and an irreversible change will occur. In addition, the recognition environment is relatively special, which is very different from the paper graphic code media in other industries. Common recognition algorithms cannot accurately recognize it well. This requires targeted research and development and optimization of image equipment algorithms dedicated to the nuclear energy field.

[0005] However, the demand for the nuclear energy industry is much smaller than that for other industries, and general suppliers do not have special testing environments and conditions such as nuclear radiation. Therefore, the image recognition technology for graphic codes such as one-dimensional codes, two-dimensional codes, and text on the surfaces of special containers in nuclear radiation environments has been developing relatively slowly. At present, there are no image recognition technology products on the market that are specifically for graphic codes such as one-dimensional codes, two-dimensional codes, and text on the surfaces of special containers in nuclear radiation environments. Summary of the invention

[0006] The technical problem to be solved by the present invention is to address the above-mentioned deficiencies in the prior art and to provide a method and device for identifying graphic codes on the surface of special containers under a nuclear radiation environment, which can realize real-time, accurate recognition and automated, intelligent and efficient recognition of graphic codes on the surface of special containers under a nuclear radiation environment.

[0007] The technical solution of the present invention to solve the above technical problems is:

[0008] According to a first aspect of the present invention, a method for identifying a graphic code on the surface of a special container in a nuclear radiation environment is provided, comprising:

[0009] (1) Obtain video stream;

[0010] (2) Performing graphic code recognition on video stream images based on a computer graphics image algorithm framework, including:

[0011] Decode the video to obtain video frames;

[0012] Select the video frame to be processed;

[0013] Perform target detection and recognition on the selected video frames to determine the target image;

[0014] Grayscale the target image to obtain a grayscale image;

[0015] Detect whether there is a graphic code in the grayscale image and determine the position of the graphic code;

[0016] According to the position of the graphic code, local cutout is performed to obtain a graphic code image;

[0017] removing noise from the graphic code image;

[0018] Performing hybrid binarization processing or OSTU binarization processing on the graphic code image after noise removal to obtain a binarized image;

[0019] Perform graphic code recognition on the binary image to obtain the recognition result;

[0020] (3) The recognition results are stored in the local storage medium and reported to the MES system.

[0021] Optionally, the selecting of the video frames to be processed specifically includes:

[0022] Count the frequency and processing speed distribution of target objects in the most recent N frames of each video;

[0023] Count the frame loss frequency of each video channel;

[0024] Based on the three characteristic parameters of the frequency of target objects appearing in the most recent N frames of each video, the processing speed distribution, and the frame loss frequency, it is determined whether each frame of each video is skipped. If not, the video frame is not skipped and is used as the video frame that needs to be processed.

[0025] Optionally, the determining whether each frame of each video stream is skipped specifically includes:

[0026] Assign weights to each feature parameter based on the priority of the strategy;

[0027] Based on the statistical feature parameters and corresponding weights, the final frame loss probability of each video channel is calculated;

[0028] According to the final frame loss probability, determine whether each frame of each video stream is skipped.

[0029] Optionally, the performing target detection and recognition on the selected video frame to determine the target image specifically includes:

[0030] Locate the position of the graphic code on each frame of the image according to the target detection algorithm;

[0031] Track the position of the graphic code in consecutive frames according to the target tracking algorithm to form a trajectory;

[0032] Frames are selected according to the image quality in the graphic code track to determine the target image.

[0033] According to a second aspect of the present invention, there is provided a device for identifying a graphic code on the surface of a special container in a nuclear radiation environment, comprising:

[0034] An acquisition module is used to acquire video streams;

[0035] The target recognition module is connected to the acquisition module and is used to perform graphic code recognition on the video stream image according to the computer graphic image algorithm framework, including:

[0036] A decoding unit, used for decoding the video to obtain a video frame;

[0037] A video frame selection unit, used for selecting video frames to be processed;

[0038] The target detection and recognition unit is used to perform target detection and recognition on the selected video frame to determine the target image;

[0039] An image grayscale unit, used for grayscale processing the target image to obtain a grayscale image;

[0040] A graphic code detection unit, used to detect whether there is a graphic code in the grayscale image and determine the position of the graphic code;

[0041] A local cutout unit, used for performing local cutout according to the position of the graphic code to obtain a graphic code image;

[0042] A Gaussian filter unit, used to remove noise from the graphic code image;

[0043] A binarization unit, used for performing mixed binarization processing or OSTU binarization processing on the graphic code image after noise removal to obtain a binarized image;

[0044] A graphic code recognition unit is used to perform graphic code recognition on the binary image to obtain a recognition result;

[0045] The recognition result management module is connected to the target recognition module and is used to store the recognition results in the local storage medium and report to the MES system.

[0046] Optionally, the video frame selection unit includes:

[0047] The first statistical subunit is used to count the frequency and processing speed distribution of the target object in the latest N frames of each video;

[0048] The second statistical subunit is used to count the frame loss frequency of each video channel;

[0049] The video frame selector determines whether each frame of each video is skipped based on the three characteristic parameters of the frequency of the target object appearing in the most recent N frames of each video, the processing speed distribution, and the frame loss frequency. If not, the video frame is not skipped and is used as the video frame that needs to be processed.

[0050] Optionally, the target detection and identification unit includes:

[0051] The detection subunit is used to locate the position of the graphic code on each frame of the image according to the target detection algorithm;

[0052] A tracking subunit, used for tracking the position of the graphic code in consecutive frames according to a target tracking algorithm to form a trajectory;

[0053] The screening subunit is used to select frames according to the image quality in the graphic code track to determine the target image.

[0054] According to a third aspect of the present invention, there is provided a device for identifying graphic codes on the surface of special containers in a nuclear radiation environment, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the above-described method for identifying graphic codes on the surface of special containers in a nuclear radiation environment.

[0055] Beneficial effects:

[0056] The method and device for recognizing graphic codes on the surface of special containers under a nuclear radiation environment of the present invention can effectively improve the contrast between the graphic codes and the surrounding container surfaces through preprocessing technologies such as graphic binarization, solve the problems of the surface of special containers under a nuclear radiation environment being dirty and unclear after being processed by processes such as burning, and the contrast between the graphic codes on the container surface and the surrounding surfaces being reduced. By optimizing and improving the existing algorithms and adopting deep learning methods, the recognition algorithms can be continuously improved and improved, so that the one-dimensional codes, two-dimensional codes, text and other graphic codes under these special conditions on the surfaces of special containers under a nuclear radiation environment can be recognized, and accurate and efficient recognition of the graphic codes on the surfaces of special containers under a nuclear radiation environment can be achieved. Finally, the recognition results are stored locally and reported to the MES system. The entire workflow can realize automatic and continuous real-time dynamic recognition, real-time monitoring, and real-time reporting. No personnel are required to be on duty throughout the process, which perfectly meets the unmanned working mode under a nuclear radiation environment, and the recognition results can be queried and traced back at any time. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flow chart of a method for identifying a graphic code on the surface of a special container in a nuclear radiation environment according to Embodiment 1 of the present invention;

[0058] Figure 2 This is a schematic diagram of a process of preprocessing a target image according to Embodiment 1 of the present invention;

[0059] Figure 3 This is a schematic diagram of the workflow of a video frame selector according to Embodiment 1 of the present invention;

[0060] Figure 4 This is a schematic diagram of the target detection and recognition process of Example 1 of the present invention;

[0061] Figure 5 This is a schematic diagram of the structure of a device for identifying graphic codes on the surface of a special container in a nuclear radiation environment according to Embodiment 2 of the present invention;

[0062] Figure 6 This is a schematic diagram of the structure of a target recognition module according to Embodiment 2 of the present invention;

[0063] Figure 7 This is a schematic diagram of the structure of a device for identifying graphic codes on the surface of special containers in a nuclear radiation environment according to Example 3 of the present invention.

[0064] In the figure: 11, acquisition module; 12, target recognition module; 121, decoding sheet;

[0065] 122. Video frame selection unit; 123. Target detection and recognition unit; 124. Image graying unit; 125. Graphic code detection unit; 126. Local cutout unit; 127. Gaussian filtering unit; 128. Binarization unit; 129. Graphic code recognition unit; 13. Recognition result management module; 21. Memory; 22. Processor. DETAILED DESCRIPTION

[0066] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0067] In the description of the present invention, it should be noted that the term "upper" and the like to indicate orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience and simplification of description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0068] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0069] In view of the fact that the graphic codes on the surface of special containers under nuclear radiation environment are greatly different from paper graphic code media in other industries, and it is difficult to accurately identify them using common recognition algorithms in the prior art, the present invention discloses a method for identifying the graphic codes on the surface of special containers under nuclear radiation environment, comprising:

[0070] (1) Obtain video stream;

[0071] (2) Performing graphic code recognition on video stream images based on a computer graphics image algorithm framework, including:

[0072] Decode the video to obtain video frames;

[0073] Select the video frame to be processed;

[0074] Perform target detection and recognition on the selected video frames to determine the target image;

[0075] Grayscale the target image to obtain a grayscale image;

[0076] Detect whether there is a graphic code in the grayscale image and determine the position of the graphic code;

[0077] According to the position of the graphic code, local cutout is performed to obtain a graphic code image;

[0078] removing noise from the graphic code image;

[0079] Performing hybrid binarization processing or OSTU (Otsu's Method) binarization processing on the graphic code image after noise removal to obtain a binarized image;

[0080] Perform graphic code recognition on the binary image to obtain the recognition result;

[0081] (3) Store the recognition results and report them to the MES system.

[0082] Accordingly, the present invention also discloses a method and device for identifying a graphic code on the surface of a special container in a nuclear radiation environment, comprising:

[0083] An acquisition module is used to acquire video streams;

[0084] The target recognition module is connected to the acquisition module and is used to perform graphic code recognition on the video stream image according to the computer graphic image algorithm framework, including:

[0085] A decoding unit, used for decoding the video to obtain a video frame;

[0086] A video frame selection unit, used for selecting video frames to be processed;

[0087] The target detection and recognition unit is used to perform target detection and recognition on the selected video frame to determine the target image;

[0088] An image grayscale unit, used for grayscale processing the target image to obtain a grayscale image;

[0089] A graphic code detection unit, used to detect whether there is a graphic code in the grayscale image and determine the position of the graphic code;

[0090] A local cutout unit, used for performing local cutout according to the position of the graphic code to obtain a graphic code image;

[0091] A Gaussian filter unit, used to remove noise from the graphic code image;

[0092] A binarization unit, used for performing mixed binarization processing or OSTU binarization processing on the graphic code image after noise removal to obtain a binarized image;

[0093] A graphic code recognition unit is used to perform graphic code recognition on the binary image to obtain a recognition result;

[0094] The recognition result management module is connected to the target recognition module and is used to store the recognition results in the local storage medium and report to the MES system.

[0095] The method and device for recognizing graphic codes on the surface of special containers in a nuclear radiation environment of the present invention can realize real-time accurate recognition and automatic, intelligent and efficient recognition of graphic codes on the surface of special containers in a nuclear radiation environment.

[0096] Example 1

[0097] like Figure 1 As shown, this embodiment discloses a method for identifying a graphic code on the surface of a special container in a nuclear radiation environment, comprising:

[0098] Step (1) obtains the video stream. Specifically, the video stream data can be obtained from a camera, a video file, or a network stream.

[0099] Step (2) performs graphic code recognition on the video stream image according to the computer graphic image algorithm framework. Taking the video stream obtained from the camera as an example, the workstation camera is configured to collect images, and the video stream of the workstation camera is connected to the host computer. The host computer performs video decoding, graphic code detection, graphic code recognition and other tasks, including:

[0100] Decode the video to obtain video frames;

[0101] Select the video frame to be processed through the video frame selector;

[0102] Perform target detection and recognition on the selected video frames to determine the target image;

[0103] like Figure 2 As shown, the target image is preprocessed, specifically including: graying the target image, converting the target image from RGB three channels to a single channel to obtain a grayscale image;

[0104] Detect whether there is a graphic code in the grayscale image, that is, Figure 2 The barcode detection shown determines the position of the graphic code. In this embodiment, a deep neural network model is specifically used to detect whether there is a graphic code in the grayscale image;

[0105] According to the position of the graphic code, the diffusion is performed according to a certain coefficient. In this embodiment, the diffusion coefficient is dynamically adjusted according to the distance from the position of the graphic code to the boundary. The general value range is [1.1, 1.5]. Local cutout is performed to cut out the area separately to obtain the graphic code image, so as to reduce the interference of the stain area;

[0106] Gaussian filtering to remove the fine noise of the graphic code image (specifically refers to the distortion caused by illumination, angle, movement, etc. during shooting);

[0107] The graphic code image after noise removal is subjected to mixed binarization processing or OSTU binarization processing to obtain a binarized image (i.e. Figure 4 The image preprocessing flow shown in FIG. 1 is shown in FIG. 1 , wherein the hybrid binarization process is to use the prior knowledge that the main color components of the graphic code area are black and the background color, to count the grayscale distribution of each graphic code pixel, to calculate the local optimal binarization threshold, and to convert the grayscale image into a binary image through the local binarization threshold; the OSTU binarization process is to divide the image into a foreground (target) and a background by finding an optimal threshold, so that the inter-class variance between the foreground and the background is maximized, so that the difference between the foreground and the background can be maximized, thereby obtaining a better binarization effect and a binary image;

[0108] Perform graphic code recognition on the binary image to obtain the recognition result.

[0109] Step (3) stores the recognition results in a local storage medium (database) and reports to the MES system (Manufacturing Execution System).

[0110] Considering that a host computer may analyze the video streams of multiple cameras, the number of cameras connected to it will change dynamically. At the same time, the amount of calculation required to analyze one camera will change with the frequency of appearance of the target object. Therefore, the computing load of the host computer changes dynamically. When there are many video channels connected, in order to ensure the real-time performance of video processing, the graphic code recognition program of the host computer will skip some video frames. When the frequency of appearance of the target object in a certain video channel is greater, the computing resources of the host computer will be tilted towards this video channel, that is, the video frames of this video channel will be selected as the video frames that need to be processed. Therefore, in some embodiments, the video frames that need to be processed are selected, such as Figure 3 As shown, specifically including:

[0111] Count the frequency and processing speed distribution of target objects in the most recent N frames of each accessed video;

[0112] Count the frame loss frequency of each video channel;

[0113] Based on the three characteristic parameters of the frequency of target objects appearing in the most recent N frames of each video, the processing speed distribution, and the frame loss frequency, it is determined whether each frame of each video is skipped, wherein: if so, the video frame is skipped, and when performing graphic code recognition, the graphic code recognition program directly discards the video frame and obtains the next frame from the video decoder; if not, the video frame is not skipped and is used as the video frame that needs to be processed.

[0114] Specifically, the decision basis for selecting the video frames to be processed includes the following three characteristic parameters: the frequency of the target object appearing in each video; the processing speed distribution of each target object in each video, in other words, the average processing time of each target object in each video; and the frame loss frequency of each video.

[0115] When the frequency of target objects appearing in a certain video is high, for example, when the frequency reaches >0.2 / frame, the frames of this video will be discarded with a lower probability; when the average processing time is long, for example, when the video is 25 frames / second, when the average processing time per frame is higher than 1 / 25 second, that is, the processing speed is slow, the frames of this video will be discarded with a higher probability; when the frame loss frequency of a certain video is too large and exceeds a certain threshold, for example, when the frame loss frequency is greater than 5 frames / second, the processing frequency of the video frames will be increased with a certain probability to improve processing fairness.

[0116] When selecting the video frames to be processed, the above-mentioned feature parameters are combined with different weights as a selection strategy to determine the final frame loss probability of each video. Specifically, it is determined whether each frame of each video stream is skipped, which includes:

[0117] Assign weights to each feature parameter based on the priority of the strategy;

[0118] Based on the statistically obtained feature parameters and corresponding weights, the final frame loss probability of each video channel is calculated. In this embodiment, a machine learning method is adopted, and the frame loss probability is independently determined by a model (for example, a decision tree model, but not limited thereto, and other regression models may also be used);

[0119] According to the final frame loss probability, determine whether each frame of each video stream is skipped.

[0120] In this embodiment, the selection strategy of combining the above-mentioned feature parameters with different weights is specifically implemented through a machine learning method. Specifically, the weights are iteratively updated according to the direction of gradient descent of the loss function until a global optimal or local optimal weight combination is obtained.

[0121] It should be noted that, through the above selection strategy, multiple channels of video can be selected for frame selection, thereby supporting the simultaneous recognition of multiple cameras. In this embodiment, up to 8 channels are supported. Of course, one channel of video is also selected for frame selection, not limited to multiple channels of video, and frames with better picture quality are selected for target detection and recognition. In addition, this method can adjust the frame selection strategy according to the video load dynamics, and can adjust the image binarization threshold according to the image quality and lighting of each frame to improve the recognition rate of graphic codes.

[0122] In some embodiments, the target detection and recognition is performed on the selected video frame to determine the target image, such as Figure 4 As shown, specifically including:

[0123] Locate the position of the graphic code on each frame of the image according to a target detection algorithm (such as yolov5, SSD and other algorithms, but not limited thereto);

[0124] Tracking the position of the graphic code in consecutive frames according to a target tracking algorithm (such as, but not limited to, deepsort or Kalman filter algorithms) to form a trajectory;

[0125] Frames are selected according to the image quality in the graphic code trajectory to determine the target image. In this implementation, several frames with better image quality are selected as target images. Specifically, a deep quality model is used to judge the image quality. When the quality score is greater than a certain threshold, it is considered to be a high-quality image. The specific threshold depends on the specific application scenario and will not be described here.

[0126] In this embodiment, the graphic code includes a one-dimensional code, a two-dimensional code, and text, etc. According to the type of the graphic code, the target detection includes a one-dimensional code detection, a two-dimensional code detection, and text detection, etc. Correspondingly, the graphic code recognition includes a one-dimensional code recognition, a two-dimensional code recognition, and text recognition, etc. Whether the recognition process is terminated is determined by judging whether the target trajectory is terminated. When the target trajectory is terminated, the recognition result will be output. When the target trajectory is not terminated, the recognition result will be cached until the target trajectory is terminated and then output.

[0127] Specifically, the detection algorithm of the one-dimensional code and the two-dimensional code in this embodiment adopts the YOLO algorithm, specifically the YOLOv7 algorithm. The YOLO algorithm based on deep learning has the advantages of fast detection speed and high detection accuracy. The text detection algorithm can adopt the paddleOCR algorithm.

[0128] In some implementations, in order to improve the recognition effect, the method uses a voting method for several recognition results of the same target object to determine the final recognition result, that is, the recognition result is voted and output.

[0129] The method for recognizing graphic codes on the surfaces of special containers under a nuclear radiation environment in this embodiment can effectively improve the contrast between the graphic codes and the surrounding container surfaces through preprocessing technologies such as graphic binarization, and solve the problems of unclear surfaces of special containers under a nuclear radiation environment caused by burning and other process treatments, and decreased contrast between the graphic codes on the container surfaces and the surrounding surfaces. By optimizing and improving the existing algorithms and adopting deep learning methods, the recognition algorithms can be continuously improved and improved, so that the one-dimensional codes, two-dimensional codes, text and other graphic codes under these special conditions on the surfaces of special containers under a nuclear radiation environment can be recognized, and accurate and efficient recognition of the graphic codes on the surfaces of special containers under a nuclear radiation environment can be achieved. Finally, the recognition results are stored locally and reported to the MES system. The entire workflow can realize automatic and continuous real-time dynamic recognition, real-time monitoring, and real-time reporting. No personnel are required on duty throughout the process, which perfectly meets the unmanned working mode under a nuclear radiation environment, and the recognition results can be queried and traced back at any time.

[0130] Example 2

[0131] like Figure 5 , Figure 6 As shown, this embodiment discloses a device for identifying a graphic code on the surface of a special container in a nuclear radiation environment, which is used in the identification method described in Example 1, and includes:

[0132] An acquisition module 11 is used to acquire a video stream;

[0133] The target recognition module 12 is connected to the acquisition module 11 and is used to perform graphic code recognition on the video stream image according to the computer graphic image algorithm framework, including:

[0134] A decoding unit 121, used for decoding the video to obtain a video frame;

[0135] A video frame selection unit 122, used to select a video frame to be processed;

[0136] The target detection and recognition unit 123 is used to perform target detection and recognition on the selected video frame to determine the target image;

[0137] An image grayscale unit 124 is used to grayscale the target image to obtain a grayscale image;

[0138] A graphic code detection unit 124, used to detect whether there is a graphic code in the grayscale image and determine the position of the graphic code;

[0139] A local cutout unit 126, used to perform local cutout according to the position of the graphic code to obtain a graphic code image;

[0140] A Gaussian filter unit 127, used to remove noise from the graphic code image;

[0141] A binarization unit 128, used for performing a mixed binarization process or an OSTU binarization process on the graphic code image after noise removal to obtain a binarized image;

[0142] The graphic code recognition unit 129 is used to perform graphic code recognition on the binary image to obtain a recognition result;

[0143] The recognition result management module 13 is connected to the target recognition module 12, and specifically includes a recognition result storage unit and a recognition result reporting unit (such as Figure 1 As shown), it is used to store the recognition results in the local storage medium (database) and report to the MES system.

[0144] In this embodiment, the acquisition module 11 may specifically be a camera management module, which acquires video stream data through camera collection.

[0145] In this embodiment, Figure 1 As shown, the graphic code recognition unit 129 is provided with a one-dimensional code recognition algorithm, a two-dimensional code recognition algorithm and a text recognition algorithm, so that the one-dimensional code, the two-dimensional code and the text can be recognized at the same time.

[0146] In some implementations, the video frame selection unit 122 includes:

[0147] The first statistical subunit is used to count the frequency and processing speed distribution of the target object in the latest N frames of each video;

[0148] The second statistical subunit is used to count the frame loss frequency of each video channel;

[0149] The video frame selector determines whether each frame of each video is skipped based on the three characteristic parameters of the frequency of the target object appearing in the most recent N frames of each video, the processing speed distribution, and the frame loss frequency. If not, the video frame is not skipped and is used as the video frame that needs to be processed. If so, the video frame is skipped. When performing graphic code recognition, the graphic code recognition program directly discards the video frame and obtains the next frame from the video decoder.

[0150] In some implementations, the target detection and identification unit 123 includes:

[0151] The detection subunit is used to locate the position of the graphic code on each frame of the image according to the target detection algorithm;

[0152] A tracking subunit, used for tracking the position of the graphic code in consecutive frames according to a target tracking algorithm to form a trajectory;

[0153] The screening subunit is used to select frames according to the image quality in the graphic code track to determine the target image.

[0154] Example 3

[0155] like Figure 7 As shown, this embodiment discloses a device for identifying graphic codes on the surface of special containers in a nuclear radiation environment, which includes a memory 21 and a processor 22, wherein the memory 21 stores a computer program, and the processor 22 is configured to run the computer program to execute the method for identifying graphic codes on the surface of special containers in a nuclear radiation environment described in Example 1.

[0156] The memory 21 is connected to the processor 22. The memory 21 may be a flash memory, a read-only memory or other memory. The processor 22 may be a central processing unit or a single-chip microcomputer.

[0157] The device for recognizing graphic codes on the surface of special containers under a nuclear radiation environment disclosed in Examples 2 to 3 can effectively improve the contrast between the graphic codes and the surrounding container surfaces through preprocessing technologies such as graphic binarization, solve the problems of unclear surfaces of special containers under a nuclear radiation environment caused by burning and other process treatments, and decreased contrast between the graphic codes on the container surfaces and the surrounding surfaces. By optimizing and improving the existing algorithms and adopting deep learning methods, the recognition algorithms can be continuously improved and improved, so that the one-dimensional codes, two-dimensional codes, text and other graphic codes under these special conditions on the surfaces of special containers under a nuclear radiation environment can be recognized, and accurate and efficient recognition of the graphic codes on the surfaces of special containers under a nuclear radiation environment can be achieved. Finally, the recognition results are stored locally and reported to the MES system. The entire workflow can realize automatic and continuous real-time dynamic recognition, real-time monitoring, and real-time reporting. No personnel are required on duty throughout the process, which perfectly meets the unmanned working mode under a nuclear radiation environment, and the recognition results can be queried and traced back at any time.

[0158] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for identifying graphic codes on the surface of special containers in a nuclear radiation environment, characterized in that: include: (1) Obtain video stream; (2) Performing graphic code recognition on the video stream image based on the computer graphics image algorithm framework, including: Decode the video to obtain video frames; Select the video frame to be processed; Perform target detection and recognition on the selected video frames to determine the target image; Grayscale the target image to obtain a grayscale image; Detect whether there is a graphic code in the grayscale image and determine the position of the graphic code; According to the position of the graphic code, local cutout is performed to obtain a graphic code image; removing noise from the graphic code image; Performing hybrid binarization processing or OSTU binarization processing on the graphic code image after noise removal to obtain a binarized image; Perform graphic code recognition on the binary image to obtain the recognition result; (3) The recognition results are stored in the local storage medium and reported to the MES system.

2. The method for identifying graphic codes on the surface of special containers in a nuclear radiation environment according to claim 1, characterized in that: The selecting of the video frames to be processed specifically includes: Count the frequency and processing speed distribution of target objects in the most recent N frames of each video; Count the frame loss frequency of each video channel; Based on the three characteristic parameters of the frequency of the target object appearing in the most recent N frames of each video, the distribution of processing speed, and the frame loss frequency, it is determined whether each frame of each video is skipped. If not, the video frame is not skipped and is treated as the video frame that needs to be processed.

3. The method for identifying the surface graphic code of a special container under a nuclear radiation environment according to claim 2 is characterized in that: The determining whether each frame of each video stream is skipped specifically includes: Assign weights to each feature parameter based on the priority of the strategy; Based on the statistical feature parameters and corresponding weights, the final frame loss probability of each video channel is calculated; According to the final frame loss probability, determine whether each frame of each video stream is skipped.

4. The method for identifying the surface graphic code of a special container in a nuclear radiation environment according to claim 1, characterized in that: The performing target detection and recognition on the selected video frame to determine the target image specifically includes: Locate the position of the graphic code on each frame of the image according to the target detection algorithm; Track the position of the graphic code in consecutive frames according to the target tracking algorithm to form a trajectory; Frames are selected according to the image quality in the graphic code track to determine the target image.

5. A device for identifying graphic codes on the surface of special containers in a nuclear radiation environment, characterized in that: include: An acquisition module (11), used for acquiring a video stream; The target recognition module (12) is connected to the acquisition module and is used to perform graphic code recognition on the video stream image according to the computer graphic image algorithm framework, including: A decoding unit (121), used for decoding the video to obtain a video frame; A video frame selection unit (122), used for selecting a video frame to be processed; A target detection and recognition unit (123) is used to perform target detection and recognition on the selected video frame to determine the target image; An image grayscale unit (124), used for grayscale processing the target image to obtain a grayscale image; A graphic code detection unit (125), used to detect whether a graphic code exists in the grayscale image and determine the position of the graphic code; A local cutout unit (126) is used to perform local cutout according to the position of the graphic code to obtain a graphic code image; A Gaussian filter unit (127), used for removing noise from the graphic code image; A binarization unit (128) is used to perform a mixed binarization process or an OSTU binarization process on the graphic code image after the noise is removed to obtain a binarized image; A graphic code recognition unit (129) is used to perform graphic code recognition on the binary image to obtain a recognition result; The recognition result management module (13) is connected to the target recognition module and is used to store the recognition result in a local storage medium and report it to the MES system.

6. The device for identifying graphic codes on the surface of special containers in a nuclear radiation environment according to claim 5, characterized in that: The video frame selection unit comprises: The first statistical subunit is used to count the frequency and processing speed distribution of the target object in the latest N frames of each video; The second statistical subunit is used to count the frame loss frequency of each video channel; The video frame selector determines whether to skip each frame of each video channel based on the three characteristic parameters of the frequency of the target object appearing in the most recent N frames of each video channel, the distribution of processing speed, and the frame loss frequency. If not, the video frame is not skipped and is treated as the video frame that needs to be processed.

7. The device for identifying the surface graphic code of a special container in a nuclear radiation environment according to claim 5, characterized in that: The target detection and identification unit comprises: The detection subunit is used to locate the position of the graphic code on each frame of the image according to the target detection algorithm; A tracking subunit, used for tracking the position of the graphic code in consecutive frames according to a target tracking algorithm to form a trajectory; The screening subunit is used to select frames according to the image quality in the graphic code track to determine the target image.

8. A device for identifying graphic codes on the surface of special containers in a nuclear radiation environment, characterized in that: It comprises a memory (21) and a processor (22), wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the method for recognizing a surface graphic code of a special container in a nuclear radiation environment according to any one of claims 1 to 4.

Images