A static load identification system and identification method based on machine vision

Abstract

The present invention relates to a static load stacking identification system based on machine vision, including a test platform, a background service end and a mobile end, wherein the mobile end includes an image acquisition module, a GPS positioning module, an accelerometer, a compass and a data receiving and viewing module that are interconnected, and utilizes machine vision and deep learning technology to achieve consistency identification of the static load stacking test platform by comparing multiple acquired images, and to determine whether the test platform has been moved or changed or whether there is cheating in the test. The present invention has the advantages of easy operation, improved identification efficiency, reduced cost, improved intelligence level, and synchronous identification of multiple mobile terminals and multiple test platforms.

Description

Technical Field

[0001] The present invention relates to the technical field of static load heap identification, and in particular to a static load heap identification system and an identification method thereof based on machine vision. Background Art

[0002] In construction projects, the bearing capacity of the foundation must be experimentally analyzed. The static load test measurement method is currently widely used. Among them, the reliability of static load test data is an important guarantee for engineering safety. At this time, real-time monitoring of the experimental platform is a very critical link, which ensures the reliability of the experimental data.

[0003] Currently, the existing overload test borrows from the previous test, which is a separate manual operation with low efficiency. It can only be tested one-on-one, and there is cheating. It cannot keep up with the development requirements of the intelligent era.

[0004] To this end, we have developed a static load identification system and identification method based on machine vision to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and to provide a static load identification system and identification method based on machine vision, which has the advantages of easy operation, improved identification efficiency, reduced cost, improved intelligence level, and simultaneous identification of multiple mobile terminals and multiple test platforms.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a static load stacking identification system based on machine vision, including a test platform, a background server and a mobile terminal, the mobile terminal including an interconnected image acquisition module, a GPS positioning module, an accelerometer, a compass and a data receiving and viewing module, the background server including an interconnected cloud computing service module and a background data service module, a stacking is provided at the top of the test platform, the background server and the mobile terminal are connected via the Internet, the image acquisition module collects the orientation information of the four side images of the test platform, and then the background server performs data analysis and comparison, and the identification results are sent to the mobile terminal.

[0007] Preferably, a recognition method of a static load recognition system based on machine vision comprises the following steps:

[0008] 2.1. When the test platform including the stacked load is completed, the mobile terminal is used to collect image information of the initial states of the four surfaces of the test platform;

[0009] 2.2. According to the GPS positioning module, the compass and the accelerometer in the mobile terminal, the orientation information of each image is obtained, and the orientation information is bound to the image information of the initial state;

[0010] 2.3. Upload the image information of the initial state of the test platform to the backend server and save it;

[0011] 2.4. When identifying different sides of the test platform, take the image information of the four sides of the test platform again, upload it to the background server and analyze it;

[0012] 2.5. According to the set similarity threshold, determine the consistency between the image information of the initial state and the image information of the re-photographed state;

[0013] 2.6. The recognition result is saved in the background server and sent to the mobile terminal of the supervising user at the same time. The mobile terminal is used for supervising and viewing the recognition result.

[0014] Preferably, step 2.4 includes the following steps:

[0015] 2.4.1. In the analysis operation, the corresponding side image is selected according to the orientation information, and the image comparison operation is performed based on machine vision and deep learning;

[0016] 2.4.2. In the contrast operation, the image is first filtered and blurred to remove interference information in the environment;

[0017] 2.4.3. Put the color and shape features of the pile and the gaps between the piles in the image into a deep learning framework for training, identify the piles and the gaps between them, and segment and extract the specific areas of each pile and the gaps between them;

[0018] 2.4.4. Calculate the area of ​​each of the piles by combining morphological and grayscale features to obtain the contour position and area of ​​each of the piles, the number of the piles, the relative distance between the piles, and the separation line information data between the piles;

[0019] 2.4.5. According to the feature information in step 2.4.4, the corresponding side images are compared to determine the similarity of the corresponding side images.

[0020] Preferably, step 2.5 includes the following steps:

[0021] 2.5.1. If the feature similarity with the image information of the initial state is high, the recognition result of the test platform fails, the test platform is not transported and hoisted at the new pile position, and there is cheating in static load detection;

[0022] 2.5.2. If the feature similarity with the image information of the initial state is low, the test platform recognition result is passed, and the test platform is transported and hoisted to a new pile position.

[0023] Preferably, the number of the background service terminals and the mobile terminals is greater than or equal to one.

[0024] Preferably, the mobile terminal is a smart phone, the background service terminal is a background server, and a plurality of the mobile terminals can be synchronously connected to at least one background service terminal.

[0025] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art:

[0026] 1. The static load heap loading recognition system and recognition method based on machine vision of the present invention realizes accurate recognition of the static load heap loading test, ensures the reliability of the static load heap loading test, ensures that the test platform has been turned over, and prevents cheating during the static load heap loading test;

[0027] 2. The present invention is simple to operate, easy to use, improves recognition efficiency and reduces costs;

[0028] 3. The present invention combines machine vision, deep learning and cloud computing service technology to improve the level of intelligence;

[0029] 4. The present invention can realize the synchronous identification of multiple test platforms by multiple mobile terminals without interfering with each other. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Attached Figure 1 It is a structural schematic diagram of the static load pile loading identification system based on machine vision of the present invention. DETAILED DESCRIPTION

[0031] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0032] Attached Figure 1 In the invention, a static load pile identification system based on machine vision includes a test platform 10, a background service terminal 20 and a mobile terminal 30. The mobile terminal 30 includes an image acquisition module, a GPS positioning module, an accelerometer, a compass and a data receiving and viewing module which are interconnected. The background service terminal 20 includes a cloud computing service module and a background data service module which are interconnected. A pile 15 is provided at the top of the test platform 10. The background service terminal 20 and the mobile terminal 30 are connected via the Internet. The image acquisition module collects the orientation information of the four side images of the test platform 10. Then the background service terminal 20 performs data analysis and comparison, and the identification result is sent to the mobile terminal 30.

[0033] A recognition method for a static load recognition system based on machine vision comprises the following steps:

[0034] 2.1. When the test platform 10 with the pile load 15 is built, the mobile terminal 30 is used to collect image information of the initial state of the four surfaces of the test platform 10;

[0035] 2.2. According to the GPS positioning module, compass and accelerometer in the mobile terminal 30, the orientation information of each image is obtained, and the orientation information is bound to the image information of the initial state;

[0036] 2.3. Upload the image information of the initial state of the test platform 10 to the background server 20 and save it;

[0037] 2.4. When different sides of the test platform 10 are identified, the four side image information of the test platform 10 are photographed again, uploaded to the backend server 20 and analyzed; the following steps are included:

[0038] 2.4.1. In the analysis operation, the corresponding side image is selected according to the orientation information, and the image comparison operation is performed based on machine vision and deep learning;

[0039] 2.4.2. In the contrast operation, the image is first filtered and blurred to remove interference information in the environment;

[0040] 2.4.3. Put the color and shape features of the pile 15 and the gaps between the pile 15 in the image into the deep learning framework for training, identify the pile 15 and the gaps between them, and segment and extract the specific areas of each pile 15 and the gaps between them;

[0041] 2.4.4. Calculate the area of ​​each pile 15 by combining morphological and grayscale features to obtain the contour position and area of ​​each pile 15, the number of piles 15, the relative distance between the piles 15, and the separation line information data between the piles 15;

[0042] 2.4.5. According to the feature information in step 2.4.4, the corresponding side images are compared to determine the similarity of the corresponding side images.

[0043] 2.5. According to the set similarity threshold, the consistency between the image information of the initial state and the image information of the re-photographed state is determined; the following steps are included:

[0044] 2.5.1. If the feature similarity with the image information of the initial state is high, the recognition result of the test platform 10 fails, the test platform 10 is not transported and hoisted at the new pile position, and there is cheating in static load detection;

[0045] 2.5.2. If the feature similarity with the image information of the initial state is low, the recognition result of the test platform 10 is passed, and the test platform 10 is transported and hoisted to a new pile position.

[0046] 2.6. The recognition result is saved in the background server 20 and sent to the mobile terminal 30 of the supervising user at the same time. The mobile terminal 30 is used for supervising and viewing the recognition result.

[0047] The number of the backend server 20 and the number of the mobile terminals 30 is greater than or equal to one. The mobile terminal 30 is a smart phone, and the backend server 20 is a backend server. Multiple mobile terminals 30 can be connected to at least one backend server 20 simultaneously.

[0048] The present invention utilizes machine vision and deep learning technology to achieve consistency recognition of the static load test platform 10 by comparing the four side images of the test platform 10 collected multiple times, and determines whether the test platform 10 has been moved or changed, or whether there is any test cheating.

[0049] The present invention has the following advantages:

[0050] 1. The static load pile loading identification system and identification method based on machine vision described in the present invention can realize accurate identification of static load pile loading experiments, ensure the reliability of static load pile loading experiments, ensure that the test platform has been turned over, and prevent cheating during the static load pile loading test.

[0051] 2. The present invention is simple to operate, easy to use, improves recognition efficiency and reduces costs.

[0052] 3. The present invention combines machine vision, deep learning and cloud computing service technology to improve the level of intelligence.

[0053] 4. The present invention can realize the synchronous identification of multiple test platforms by multiple mobile terminals without interfering with each other.

[0054] The above are only specific application examples of the present invention and do not constitute any limitation on the protection scope of the present invention. Any technical solution formed by equivalent transformation or equivalent replacement shall fall within the protection scope of the present invention.

Claims

1. A static load identification system based on machine vision, characterized in that: It includes a test platform, a background service end and a mobile end. The mobile end includes an image acquisition module, a GPS positioning module, an accelerometer, a compass and a data receiving and viewing module that are interconnected. The background service end includes a cloud computing service module and a background data service module that are interconnected. A stack is provided at the top of the test platform. The background service end and the mobile end are connected via the Internet. The image acquisition module collects the orientation information of the four side images of the test platform. Then the background service end performs data analysis and comparison, and the recognition result is sent to the mobile end. The identification method of the static load heap load identification system based on machine vision comprises the following steps: 2.

1. When the test platform including the stacked load is completed, the mobile terminal is used to collect image information of the initial states of the four surfaces of the test platform; 2.

2. According to the GPS positioning module, the compass and the accelerometer in the mobile terminal, the orientation information of each image is obtained, and the orientation information is bound to the image information of the initial state; 2.

3. Upload the image information of the initial state of the test platform to the backend server and save it; 2.

4. When identifying different sides of the test platform, take the image information of the four sides of the test platform again, upload it to the background server and analyze it; 2.

5. According to the set similarity threshold, determine the consistency between the initial image information and the re-photographed image information: 2.5.

1. If the feature similarity with the image information of the initial state is high, the recognition result of the test platform fails, the test platform is not transported and hoisted at the new pile position, and there is cheating in static load detection; 2.5.

2. If the feature similarity with the image information of the initial state is low, the test platform recognition result is passed, and the test platform is transported and hoisted to a new pile position; 2.

6. The recognition result is saved in the background server and sent to the mobile terminal of the supervising user at the same time. The mobile terminal is used for supervising and viewing the recognition result.

2. The static load identification system based on machine vision according to claim 1 is characterized in that: Step 2.4 includes the following steps: 2.4.

1. In the analysis operation, the corresponding side image is selected according to the orientation information, and the image comparison operation is performed based on machine vision and deep learning; 2.4.

2. In the contrast operation, the image is first filtered and blurred to remove interference information in the environment; 2.4.

3. Put the color and shape features of the pile and the gaps between the piles in the image into a deep learning framework for training, identify the piles and the gaps between them, and segment and extract the specific areas of each pile and the gaps between them; 2.4.

4. Calculate the area of ​​each of the piles by combining morphological and grayscale features to obtain the contour position and area of ​​each of the piles, the number of the piles, the relative distance between the piles, and the separation line information data between the piles; 2.4.

5. According to the feature information in step 2.4.4, the corresponding side images are compared to determine the similarity of the corresponding side images.

3. The static load identification system based on machine vision according to claim 2 is characterized in that: The number of the background service terminals and the mobile terminals is greater than or equal to one.

4. The static load identification system based on machine vision according to claim 3 is characterized in that: The mobile terminal is a smart phone, the background service terminal is a background server, and a plurality of the mobile terminals can be synchronously connected to at least one background service terminal.

Images