Intelligent authentication method and system for ship

Abstract

The invention relates to the field of ship body load measurement, and particularly discloses an intelligent weight identification method and system for a ship. The camera and the high-power variable-focus unmanned aerial vehicle are used for shooting the water gauge measurement picture, the safety of outer side water gauge measurement is improved, screenshot is carried out in the shot picture at any time for water gauge draft scale detection, and a user and a supervisor observe the water gauge reading change at any time through the shot picture; the water gauge reading is intelligently judged through a draught recognition algorithm built in the rear end, the water gauge reading is transmitted back to the flat plate end, unidentified data and data with large recognition deviation are automatically filtered out, variance calculation is conducted on effective data, a stable reading is obtained, and then a displacement meter and a ballast meter of a ship are input into the system. The system calculates the average draft of the ship and the load of the ship according to the water gauge reading, the displacement meter data, the ballast water meter data and other related storage data, and the unloading weight or the loading cargo weight of the ship is obtained by measuring the load of the ship twice and calculating the difference value.

Description

Technical Field

[0001] The present invention belongs to the technical field of ship load measurement, and in particular relates to an intelligent ship load measurement method and system. Background Art

[0002] Water transportation has been popular for a long time because of its large transport weight and low price. However, for the safety of ships, the load of ships needs to be strictly controlled, because the load of ships affects the draft of ships. Operators can check the draft of ships through the water gauge on the hull. Generally, there are 6 water gauge points on the hull, distributed at the bow, the middle and the stern, and distributed on both sides of the hull. The traditional measurement method is to observe the inner water gauge through the artificial shore, and the observer will take a tugboat to the outside of the ship, observe the water gauge reading with the naked eye, and then judge the water gauge value based on experience. However, this method of observing the water gauge with the naked eye has no unified standard and is highly dependent on the experience of the operator. In addition, it is necessary to take a tugboat to observe the water gauge on the outside, which is not only time-consuming and labor-intensive, but also has a high cost for renting a ship. Moreover, it is carried out on the sea with complex conditions, which has a high risk of short-distance use. Observers standing on the shore also have certain risks due to the complex dock operations. In addition to these observation factors, the water level scales on the surface of some ships are manually painted when they leave the factory, and the font styles are different, which makes it difficult to identify them. Moreover, for ships with different loads, the water level scales at the bow and stern will have different degrees of inclination, which makes it difficult to intelligently interpret the distance values ​​between the identification scales. Moreover, after a long period of sea voyage, the water level scales on the hull will suffer from varying degrees of rust, and even directly lead to complete rust of a single scale value. In addition, the dock measurement environment is complex and there are too many interference factors, all of which will lead to difficulties in water level measurement. Summary of the invention

[0003] The purpose of the present invention is to provide a smart weight detection method and system for ships to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A smart weight detection method for ships, comprising the following steps:

[0006] S1. Set up a shore camera at the dock of the ship, and point the front of the shore camera at the water level scale inside the ship, and use a zoom camera drone to observe the water level scale outside the ship;

[0007] S2. The control terminal obtains the images taken by the drone and the camera in real time. The user and supervisor observe the changes in the water gauge reading in real time and start and stop the water gauge detection at any time;

[0008] S3. Start the load measurement at the control end, capture the pictures from the drone and camera, measure and judge the water gauge by the draft recognition algorithm at the back end to get the water gauge data, and transmit the water gauge data to the control end, which performs data screening and processing to get a final water gauge reading;

[0009] S4. Input the displacement meter and ballast water meter of the ship, and calculate the average draft and ship load of the ship according to the water gauge reading, the displacement meter and ballast water meter of the ship;

[0010] S5. Calculate and obtain the ship's load before and after loading or unloading, and calculate the difference to obtain the weight of the cargo unloaded or the weight of the cargo loaded this time.

[0011] Preferably, in S3, the draft gauge identification algorithm specifically includes the following steps:

[0012] S31, extraction and preprocessing of the target image, intercepting and averaging the video image collected by the camera, and then gray-scale conversion of the intercepted color image for subsequent processing;

[0013] S32, extracting and morphologically processing the region of interest, filtering and denoising the grayscale image, removing the noise generated by the environment or interference, retaining the image details, using edge detection on the filtered image, extracting the scale features of the water gauge, reducing the amount of data and removing unnecessary image information, retaining the image structure attributes, performing edge detection on the image and then binarizing it to remove non-target images such as the water surface and background, using morphological corrosion processing to further filter out the noise points that interfere with judgment in the binarized image, performing image refinement processing on the horizontal scale lines obtained after removing the background and the water surface through corrosion, and obtaining a complete image of the water gauge;

[0014] S33, identify the waterline, use the RG dual-channel pixel difference method or the deep learning algorithm based on Faster R-CNN to identify the waterline between the hull and the water body, and attach the waterline to the complete image of the water gauge to calculate the water gauge reading;

[0015] S34, obtaining all water gauge readings inside and outside the hull, automatically filtering unrecognized and deviated data, performing variance calculation on valid data, and obtaining a final water gauge reading.

[0016] Preferably, in S31, a video processing function provided by Microsoft's MSDN and a video frame capture function provided by OpenCV are used to extract and save multiple frames of images from the video captured by the camera, and continuous multiple frames of images are captured and averaged to avoid abnormal interference.

[0017] Preferably, in S31, when graying the color image, the conversion formula is:

[0018] T=0.299×R+0.587×G+0.114×B

[0019] Among them, T is the grayscale value of the target image after conversion, and R, D, and B are the red, green, and blue color image components respectively.

[0020] Preferably, in S32, the image denoising adopts a median filtering method, and the gray value of each pixel is reset according to the median gray value of all pixels in a certain area around the pixel, and the points that are different from the median gray value of the pixels in the area are eliminated. The specific steps include:

[0021] A1. Set a sampling window with an odd number of points;

[0022] A2. Take out an odd number of pixels from the processed image through a sampling window and sort the grayscale values ​​of these pixels;

[0023] A3. Select the median of the odd grayscale values, that is, the one in the middle from smallest to largest, as the replacement value;

[0024] A4. Set the gray value of the pixel at the center of the sampling window as the replacement value;

[0025] A5. Repeat A2-A4 to process all pixels in the image.

[0026] Preferably, in S32, an iterative threshold method is used to select a global threshold for binarization, and the algorithm flow of the iterative threshold method includes:

[0027] B1. Set a randomly generated initial threshold T 0 ;

[0028] B2. Divide the image into two parts according to the initial threshold, the target image area S 1 and background image area S 2 ;

[0029] B3. Calculate S 1 and S 2 The average grayscale value of the pixel in the part G 1 and G 2 ;

[0030] B4. Set a new threshold T n For (G 1 +G 2 ) / 2;

[0031] B5. Repeat steps B2-B4, continue to segment the image with the newly set threshold, and calculate the average gray value until the new threshold is equal to the previous threshold.

[0032] Preferably, in S32, the binary image is subjected to corrosion morphological processing, which specifically includes:

[0033] C1. Scan each pixel of the binary image X to be eroded using the structural element Y;

[0034] C2. Perform an “AND” operation on the pixel of the structural element Y and the binary image area it covers. That is, if all are 1, the pixel of the result image is 1, otherwise it is 0.

[0035] Preferably, in S32, when the image is thinned, the image skeleton is first extracted from the target image, that is, the lines with a pixel width greater than 1 in the original image are thinned into lines with a width of only 1 pixel, so as to facilitate image analysis or image feature extraction.

[0036] Preferably, in S34, when calculating the water gauge reading, the following calculation formula can be used:

[0037]

[0038] Among them, N is the reading, the key scale line array is [X1, X2, X3...Xn], the values ​​corresponding to the scale lines are recorded as [V1, V2, V3...Vn], and the water level line position is T.

[0039] A smart ship weight detection system, using a smart ship weight detection method, specifically includes:

[0040] An image acquisition system, the image acquisition system comprising a camera installed on the shore and a drone capable of taking pictures;

[0041] A processing and analysis system, the processing and analysis system is used to process the image data acquired by the image acquisition system, and analyze and process to obtain the draft reading of the water gauge in the image;

[0042] The image acquisition system is connected to the processing and analysis system via a wireless network or a wired network.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention sets a camera on the shore to shoot the water gauge scale inside the ship, thereby reducing the risk of manually observing the water gauge inside the shore. A high-power variable-focus drone is then used to measure the water gauge outside the ship at a long distance, thereby improving the safety of the water gauge measurement outside. A tablet is used to obtain shooting pictures of the drone and the camera, and screenshots are taken at any time in the shooting pictures to detect the water gauge draft scale. Users and supervisors can observe the change of the water gauge reading at any time through the shooting pictures. The tablet can also be connected to the back end, and the water gauge reading is intelligently judged by the draft recognition algorithm built into the back end, and then the water gauge reading is sent back to the tablet end, which performs data screening and automatically filters out data that is not successfully recognized and has a large recognition deviation. Finally, the effective data is calculated for variance to obtain a final reading that tends to be stable, and then the displacement meter and the ballast water meter of the ship are entered into the system. The system calculates the average draft of the ship and the ship load according to the water gauge reading, the displacement meter data, the ballast water meter data and other related stored data. By measuring the ship load twice before and after and calculating the difference, the unloading weight of the ship or the weight of the cargo loaded this time can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a structural block diagram of the intelligent weight identification method of the present invention;

[0046] Figure 2 This is a structural block diagram of the intelligent weight detection system of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1 - Figure 2 As shown, a smart weight detection method for ships specifically includes the following steps:

[0049] S1. Set up a shore camera at the dock of the ship, and point the front of the shore camera at the water level scale inside the ship, and use a zoom camera drone to observe the water level scale outside the ship;

[0050] S2. The control terminal obtains the images taken by the drone and the camera in real time. The user and supervisor observe the changes in the water gauge reading in real time and start and stop the water gauge detection at any time;

[0051] S3. Start the load measurement at the control end, capture the pictures from the drone and camera, measure and judge the water gauge by the draft recognition algorithm at the back end to get the water gauge data, and transmit the water gauge data to the control end, which performs data screening and processing to get a final water gauge reading;

[0052] S4. Input the displacement meter and ballast water meter of the ship, and calculate the average draft and ship load of the ship according to the water gauge reading, the displacement meter and ballast water meter of the ship;

[0053] S5. Calculate and obtain the ship's load before and after loading or unloading, and calculate the difference to obtain the weight of the cargo unloaded or the weight of the cargo loaded this time.

[0054] As can be seen from the above, when users and supervisors need to measure the load of a ship, they can use shore cameras and zoom drones to respectively shoot the water gauge scales on the inside and outside of the ship, which avoids the danger of manual on-site observation of the water gauge to a certain extent. Users and supervisors can watch the changes in the water gauge readings at any time through the images taken by cameras and drones to improve the credibility of the water gauge readings. Afterwards, they can upload screenshots of the captured images to the backend through a tablet, which will take screenshots of the video images, and judge the water gauge images in the screenshots according to the draft recognition algorithm to obtain the draft depth data. The tablet computer then screens the identified data, filters out data that has not been successfully identified and data with large identification deviations, and then performs variance calculation on the valid data to obtain a final reading that tends to be stable. After obtaining the final reading of the six-sided water gauge, the cargo weight is calculated, and the ship's displacement meter and ballast water meter are entered into the system. The system automatically analyzes the stored data such as the six-sided water gauge readings, displacement meter data, ballast water meter data, etc., calculates the ship's average draft and ship load, and measures the ship's load twice, calculates the difference, and obtains the ship's unloading weight or the weight of the cargo loaded this time.

[0055] In S3, the draft gauge identification algorithm specifically includes the following steps:

[0056] S31, extraction and preprocessing of the target image, intercepting and averaging the video image collected by the camera, and then gray-scale conversion of the intercepted color image for subsequent processing;

[0057] S32, extracting and morphologically processing the region of interest, filtering and denoising the grayscale image, removing the noise generated by the environment or interference, retaining the image details, using edge detection on the filtered image, extracting the scale features of the water gauge, reducing the amount of data and removing unnecessary image information, retaining the image structure attributes, performing edge detection on the image and then binarizing it to remove non-target images such as the water surface and background, using morphological corrosion processing to further filter out the noise points that interfere with judgment in the binarized image, performing image refinement processing on the horizontal scale lines obtained after removing the background and the water surface through corrosion, and obtaining a complete image of the water gauge;

[0058] S33, identify the waterline, use the RG dual-channel pixel difference method or the deep learning algorithm based on Faster R-CNN to identify the waterline between the hull and the water body, and attach the waterline to the complete image of the water gauge to calculate the water gauge reading;

[0059] S34, obtaining all water gauge readings inside and outside the hull, automatically filtering unrecognized and deviated data, performing variance calculation on valid data, and obtaining a final water gauge reading.

[0060] In the S31, a video processing function provided by Microsoft's MSDN and a video frame capture function provided by OpenCV are used to extract and save multiple frames of images from the video captured by the camera, and continuous multiple frames of images are captured and averaged to avoid abnormal interference.

[0061] In S31, when graying the color image, the conversion formula is:

[0062] T=0.299×R+0.587×G+0.114×B

[0063] Among them, T is the grayscale value of the target image after conversion, and R, D, and B are the red, green, and blue color image components respectively.

[0064] In S32, the image denoising uses a median filtering method to reset the grayscale value of each pixel according to the median grayscale value of all pixels in a certain area around the pixel, and eliminate points that are different from the median grayscale value of the pixels in the area. The specific steps include:

[0065] A1. Set a sampling window with an odd number of points;

[0066] A2. Take out an odd number of pixels from the processed image through a sampling window and sort the grayscale values ​​of these pixels;

[0067] A3. Select the median of the odd grayscale values, that is, the one in the middle from smallest to largest, as the replacement value;

[0068] A4. Set the gray value of the pixel at the center of the sampling window as the replacement value;

[0069] A5. Repeat A2-A4 to process all pixels in the image.

[0070] The size of the sampling window needs to be set appropriately. Generally, after median filtering, general noise points smaller than the sampling window area can be filtered out, while larger objects such as water gauge scales, water surface contours, etc. are almost completely unaffected by filtering.

[0071] Edge detection can also be used on the filtered image to better extract the features of the water ruler scale, greatly reduce the amount of data and eliminate unnecessary image information. At the same time, it can retain important image structure attributes to facilitate subsequent processing and identification. Commonly used operators in edge detection are Canny operator, or various improved Canny operators and Sobel operator. Here, since it is the recognition of the water ruler scale and the detail texture is not paid much attention, the Sobel operator is selected. The Sobel operator is simple to calculate and fast. It usually has directionality during detection. You can choose to perform horizontal detection or vertical detection, or both directions. Because the water ruler scale is in the horizontal direction in the image, vertical edge detection is performed here.

[0072] In S32, the iterative threshold method is used to select a global threshold for binarization, and a global threshold T is set. Then, according to the global threshold, the image data is divided into two parts: a part with a gray value greater than T and a part with a gray value less than T. The pixels of the part with a gray value greater than T are set to white, and the pixels of the part with a gray value less than T are set to black. The algorithm flow of the iterative threshold method includes:

[0073] B1. Set a randomly generated initial threshold T 0 ;

[0074] B2. Divide the image into two parts according to the initial threshold, the target image area S 1 and background image area S 2 ;

[0075] B3. Calculate S 1 and S 2 The average grayscale value of the pixel in the part G 1 and G 2 ;

[0076] B4. Set a new threshold T n For (G 1 +G 2 ) / 2;

[0077] B5. Repeat steps B2-B4, continue to segment the image with the newly set threshold, and calculate the average gray value until the new threshold is equal to the previous threshold.

[0078] In S32, the binary image is subjected to corrosion morphological processing to remove some noise points that interfere with the judgment, specifically including:

[0079] C1. Scan each pixel of the binary image X to be eroded using the structural element Y;

[0080] C2. Perform an “AND” operation on the pixel of the structural element Y and the binary image area it covers. That is, if all are 1, the pixel of the result image is 1, otherwise it is 0.

[0081] In the above S32, when performing thinning processing on the image, the image skeleton is first extracted from the target image, that is, the lines with pixel width greater than 1 in the original image are thinned into lines with a width of only 1 pixel, so as to facilitate image analysis or image feature extraction.

[0082] In S34, when calculating the water gauge reading, the following calculation formula can be used:

[0083]

[0084] Among them, N is the reading, the key scale line array is [X1, X2, X3...Xn], the values ​​corresponding to the scale lines are recorded as [V1, V2, V3...Vn], and the water level line position is T.

[0085] A smart weight detection system for ships, using a smart weight detection method for ships, specifically comprising:

[0086] An image acquisition system, the image acquisition system comprising a camera installed on the shore and a drone capable of taking pictures;

[0087] A processing and analysis system, the processing and analysis system is used to process the image data acquired by the image acquisition system, and analyze and process to obtain the draft reading of the water gauge in the image;

[0088] The image acquisition system is connected to the processing and analysis system via a wireless network or a wired network.

[0089] When measuring, the image acquisition system is used to collect images of multiple water gauges on the surface of the ship. The water gauge recognition algorithm in the processing and analysis system is used to identify the water gauge readings, and the multiple water gauge readings are selected. The variance calculation is performed on the water gauge readings that meet the standards. The reading obtained by the variance calculation is then used as the actual water gauge reading. The water gauge reading, ship displacement meter, ballast water meter and other related data are combined for calculation to obtain the ship's average draft and ship load.

[0090] In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other.

Claims

1. A smart weight detection method for ships, characterized in that: The specific steps include: S1. Set up a shore camera at the dock of the ship, and point the front of the shore camera at the water level scale inside the ship, and use a zoom camera drone to observe the water level scale outside the ship; S2. The control terminal obtains the images taken by the drone and the camera in real time. The user and supervisor observe the changes in the water gauge reading in real time and start and stop the water gauge detection at any time; S3. Start the load measurement at the control end, capture the pictures from the drone and camera, measure and judge the water gauge by the draft recognition algorithm at the back end to get the water gauge data, and transmit the water gauge data to the control end, which performs data screening and processing to get a final water gauge reading; S4. Input the displacement meter and ballast water meter of the ship, and calculate the average draft and ship load of the ship according to the water gauge reading, the displacement meter and ballast water meter of the ship; S5. Calculate and obtain the ship's load before and after loading or unloading, and calculate the difference to obtain the weight of the cargo unloaded or the weight of the cargo loaded this time.

2. A smart weight detection method for ships according to claim 1, characterized in that: In S3, the draft gauge identification algorithm specifically includes the following steps: S31, extracting and preprocessing the target image, intercepting and averaging the video image collected by the camera, and then gray-converting the intercepted color image for subsequent processing; S32, extracting and morphologically processing the region of interest, filtering and denoising the grayscale image, removing the noise generated by the environment or interference, retaining the image details, using edge detection on the filtered image, extracting the scale features of the water gauge, reducing the amount of data and removing unnecessary image information, retaining the image structure attributes, performing edge detection on the image and then binarizing it to remove non-target images such as the water surface and background, using morphological corrosion processing to further filter out the noise points that interfere with judgment in the binarized image, performing image refinement processing on the horizontal scale lines obtained after removing the background and the water surface through corrosion, and obtaining a complete image of the water gauge; S33, identify the waterline, use the RG dual-channel pixel difference method or the deep learning algorithm based on Faster R-CNN to identify the waterline between the hull and the water body, and attach the waterline to the complete image of the water gauge to calculate the water gauge reading; S34, obtaining all water gauge readings inside and outside the hull, automatically filtering unrecognized and deviated data, performing variance calculation on valid data, and obtaining a final water gauge reading.

3. A smart weight detection method for ships according to claim 2, characterized in that: In the S31, a video processing function provided by Microsoft's MSDN and a video frame capture function provided by OpenCV are used to extract and save multiple frames of images from the video captured by the camera, and continuous multiple frames of images are captured and averaged to avoid abnormal interference.

4. A smart weight detection method for ships according to claim 2, characterized in that: In S31, when graying the color image, the conversion formula is: T=0.299×R+0.587×G+0.114×B Among them, T is the grayscale value of the target image after conversion, and R, D, and B are the red, green, and blue color image components respectively.

5. The intelligent weight detection method for ships according to claim 2, characterized in that: In S32, the image denoising uses a median filtering method to reset the grayscale value of each pixel according to the median grayscale value of all pixels in a certain area around the pixel, and eliminate points that are different from the median grayscale value of the pixels in the area. The specific steps include: A1. Set a sampling window with an odd number of points; A2. Take out an odd number of pixels from the processed image through a sampling window and sort the grayscale values ​​of these pixels; A3. Select the median of the odd grayscale values, that is, the one in the middle from smallest to largest, as the replacement value; A4. Set the gray value of the pixel at the center of the sampling window as the replacement value; A5. Repeat A2-A4 to process all pixels in the image.

6. The intelligent weight detection method for ships according to claim 2, characterized in that: In S32, an iterative threshold method is used to select a global threshold for binarization. The algorithm flow of the iterative threshold method includes: B1. Set a randomly generated initial threshold T0; B2, dividing the image into two parts according to the initial threshold, the target image area S1 and the background image area S2; B3, respectively calculate the average grayscale values ​​G1 and G2 of the pixels in parts S1 and S2; B4. Set a new threshold T n =(G1+G2) / 2; B5. Repeat steps B2-B4, continue to segment the image with the newly set threshold, and calculate the average gray value until the new threshold is equal to the previous threshold.

7. The intelligent weight detection method for ships according to claim 2, characterized in that: In the step S32, the binary image is subjected to corrosion morphological processing, which specifically includes: C1. Scan each pixel of the binary image X to be eroded using the structural element Y; C2. Perform an "AND" operation on the pixel of the structural element Y and the binary image area it covers. That is, if all are 1, the pixel of the result image is 1, otherwise it is 0.

8. The intelligent weight detection method for ships according to claim 2, characterized in that: In the above S32, when performing thinning processing on the image, the image skeleton is first extracted from the target image, that is, the lines with a pixel width greater than 1 in the original image are thinned into lines with a width of only 1 pixel, so as to facilitate image analysis or image feature extraction.

9. The intelligent weight detection method for ships according to claim 2, characterized in that: In S34, when calculating the water gauge reading, the following calculation formula can be used: Among them, N is the reading, the key scale line array is [X1, X2, X3...Xn], the values ​​corresponding to the scale lines are recorded as [V1, V2, V3...Vn], and the water level line position is T.

10. A smart ship weight detection system, using a smart ship weight detection method according to any one of claims 1 to 9, characterized in that: Specifically include: An image acquisition system, the image acquisition system comprising a camera installed on the shore and a drone capable of taking pictures; A processing and analysis system, the processing and analysis system is used to process the image data acquired by the image acquisition system, and analyze and process to obtain the draft reading of the water gauge in the image; The image acquisition system is connected to the processing and analysis system via a wireless network or a wired network.

Images