Tank appearance defect detection device based on deep learning and 360-degree surrounding imaging

Abstract

The invention discloses a tank appearance defect detection device based on deep learning and 360-degree surrounding imaging, and belongs to the technical field of defect detection.The tank appearance defect detection device comprises a frame body, a detection assembly is detachably connected to the interior of the frame body, an eliminating tank receiving groove is further connected to the interior of the frame body, and a conveying belt is located below the detection assembly; a tank separating mechanism is arranged at the end, close to the detection assembly, of the conveying belt, the side edge of the conveying belt abuts against one end of the removing and tank receiving groove, the directions of the conveying belt and the removing and tank receiving groove are perpendicular to each other and located on the same horizontal plane, and a tank removing mechanism is fixedly installed on the side, away from the removing and tank receiving groove, of the conveying belt; the frame body is also provided with a main control center; the detection assembly is arranged, the situation that a plurality of industrial cameras are used for recording and shooting at the same time can be reduced, the 360-degree appearance state of the tank can be shot through one picture, therefore, the appearance of the tank can be detected more conveniently, the detection period is shorter, the working efficiency is higher, and the cost is lower.

Description

Technical Field

[0001] The present invention belongs to the technical field of defect detection, and in particular relates to a tank appearance defect detection device based on deep learning and 360-degree surround imaging. Background Art

[0002] At present, in the industry, especially for 360-degree no-dead-angle detection around the entire tank, the traditional method usually uses 3 to 4 industrial cameras to achieve surround shooting, which is costly and slow to process. A large number of image processing rules need to be set during image processing, and it is easy to miss defective products, resulting in defective products flowing into the market, generating complaints, and seriously affecting the brand image. Therefore, how to provide a tank appearance defect detection device based on deep learning and 360-degree surround imaging is a problem that technicians in this field urgently need to solve. Summary of the invention

[0003] The main purpose of the present invention is to provide a tank appearance defect detection device based on deep learning and 360-degree surround imaging to solve the above technical problems. The device is provided with a detection component, which can avoid using multiple industrial cameras for recording and shooting at the same time, and can take a photo to capture the appearance stitching state of the tank in different directions, making it more convenient to detect the appearance of the tank, shorten the detection cycle, and increase the work efficiency.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A can appearance defect detection device based on deep learning and 360-degree surround imaging, comprising a frame, a detection component detachably connected to the inside of the frame, a conveyor belt and a can rejection receiving slot also connected to the inside of the frame, the conveyor belt is located below the detection component, and a can separation mechanism is provided at one end of the conveyor belt close to the detection component, the side of the conveyor belt abuts against one end of the can rejection receiving slot and the directions are perpendicular to each other and located on the same horizontal plane, a can rejection mechanism is fixedly installed on the side of the conveyor belt away from the can rejection receiving slot, a can is arranged on the conveyor belt, and a main control center is also arranged on the frame, the main control center is electrically connected to the detection component, the detection component can detect defects of the can body, and the can separation mechanism can separate adjacent can bodies to a suitable distance.

[0006] Furthermore, the detection component includes an industrial camera, a fixing rod, a reflector fixing bracket and a reflective lens. A plurality of fixing rods are fixedly arranged on the inner walls on both sides of the frame close to the conveyor belt. The end of the fixing rod away from the frame is connected to the reflector fixing bracket. The reflector fixing brackets are correspondingly distributed on both sides above the conveyor belt. The reflector fixing brackets are fixedly connected to the reflective lenses. The industrial camera is fixed at the top of the frame.

[0007] Furthermore, the can rejection mechanism includes a rejection cylinder connecting frame and a rejection cylinder. The rejection cylinder connecting frame is fixed to the side of the conveyor belt away from the rejection can receiving slot. The rejection cylinder is arranged on the rejection cylinder connecting frame, and the rejection cylinder is electrically connected to the main control center.

[0008] Furthermore, the tank separation mechanism includes a tank separation sensor, a tank separation push rod support frame and a tank separation push rod. The tank separation push rod support frame is fixed on the side wall of the conveyor belt. The tank separation push rod and the tank separation sensor are connected to the tank separation push rod support frame. The tank separation push rod is a pneumatic push rod. The tank separation push rod and the tank separation sensor are both electrically connected to the main control center.

[0009] Furthermore, an industrial light source is provided inside the frame above the conveyor belt, and the industrial light source is electrically connected to the main control center.

[0010] Compared with the prior art, the present invention has the following beneficial effects:

[0011] The present invention is provided with a detection component, which can avoid using multiple industrial cameras for recording and shooting at the same time, and can capture the appearance stitching state of the tank body in different directions through one photo, so that the appearance of the tank body is more convenient to detect, the detection cycle is shorter, and the work efficiency is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0013] Figure 1 It is a schematic diagram of the structure of the present invention.

[0014] Figure 2 It is a schematic diagram of the local structure of location A of the present invention.

[0015] Figure 3 It is an isometric structural schematic diagram of the present invention.

[0016] Figure 4 Schematic diagram of the distribution position of reflective lenses.

[0017] Among them, 1- conveyor belt, 2- mirror fixing bracket, 3- tank body, 4- fixing rod, 5- reflective lens, 6- frame body, 7- rejecting cylinder, 8- rejecting can receiving slot, 9- can separation push rod, 10- industrial camera. DETAILED DESCRIPTION

[0018] 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.

[0019] like Figures 1 to 4 As shown, the present invention provides a can appearance defect detection device based on deep learning and 360-degree surround imaging, including a frame 6, wherein the frame 6 is detachably connected to a detection component, the frame 6 is also connected to a conveyor belt 1 and a can rejection receiving groove 8, the conveyor belt 1 is located below the detection component, and a can separation mechanism is provided at one end of the conveyor belt 1 close to the detection component, the side of the conveyor belt 1 abuts against one end of the can rejection receiving groove 8 and the directions are perpendicular to each other and located on the same horizontal plane, a can rejection mechanism is fixedly installed on the side of the conveyor belt 1 away from the can rejection receiving groove 8, a can 3 is provided on the conveyor belt 1 and can drive the can 3 to run at a uniform speed, a main control center is also provided on the frame 6, the main control center is electrically connected to the detection component, the detection component can detect defects of the can 3, and the can separation mechanism can separate adjacent cans 3 to a suitable distance.

[0020] In this embodiment, the detection component includes an industrial camera 10, a fixing rod 4, a reflector fixing bracket and a reflective lens 5. A plurality of fixing rods 4 are fixedly arranged on the inner walls on both sides of the frame 6 close to the conveyor belt 1. The end of the fixing rod away from the frame 6 is connected to the reflector fixing bracket. The reflector fixing brackets are distributed on both sides above the conveyor belt 1 corresponding to each other. The reflector fixing brackets are fixedly connected to the reflective lens 5. The industrial camera 10 is fixed at the top of the frame 6. The industrial camera 10 is used to assist in shooting and recording the appearance of the tank body 3 in the reflective lens 5, such as Figure 4 The reflective lens 5 shown is arranged around the tank body 3, and the reflective lens 5 is inclined at a certain angle to the horizontal plane, which can ensure that the industrial camera 10 can record the appearance of the stitching of the entire tank body 3 from the reflective lens 5, which is convenient for subsequent analysis and processing.

[0021] In this embodiment, the can body rejection mechanism includes a rejection cylinder connecting frame and a rejection cylinder 7. The rejection cylinder connecting frame is fixed to the side of the conveyor belt 1 away from the rejection can receiving groove 8. The rejection cylinder 7 is arranged on the rejection cylinder connecting frame. The rejection cylinder 7 is electrically connected to the main control center. The rejection cylinder 7 is used to push the can body 3. One end can be provided with an arc-shaped structural block, which can ensure that the can body 3 is stably pushed to the rejection can receiving groove 8 during pushing. The rejection can receiving groove 8 is used to receive the rejected can body to prevent the can body from falling.

[0022] In this embodiment, the tank separation mechanism includes a tank separation sensor, a tank separation push rod support frame and a tank separation push rod 9. The tank separation push rod support frame is fixed on the side wall of the conveyor belt 1. The tank separation push rod 9 and the tank separation sensor are connected to the tank separation push rod support frame. The tank separation push rod 9 is a pneumatic push rod. The tank separation push rod 9 and the tank separation sensor are electrically connected to the main control center. The tank separation sensor is mainly used to sense the position of the tank body 3 on the conveyor belt 1. The tank separation push rod 9 can be a pneumatic push rod, an electric hydraulic push rod or other structures, which is mainly used to block the movement of the tank body 3 on the conveyor belt 1 to achieve effective separation of the tank body intervals, with high cost performance and simple operation; and a triangular tank separation structure block can also be provided at one end of the tank separation push rod 9 used for tank separation, so that the tank separation push rod 9 can be inserted between two tank bodies 3 after being extended for tank separation.

[0023] In this embodiment, an industrial light source is also provided inside the frame 6 above the conveyor belt 1, and the industrial light source is electrically connected to the main control center. An industrial light source is also provided inside the frame 6 above the conveyor belt 1, and the industrial light source is electrically connected to the main control center. The industrial light source can be a ring-shaped lamp for auxiliary lighting, so that the illumination of the tank body is more uniform.

[0024] Working principle: For the appearance defect detection scenario of cylindrical or nearly cylindrical cans or bottles, defect data is first generated through AIGC, and pictures are acquired with the help of a low-cost 360-degree surround imaging system. Then, a deep learning algorithm is used to realize automatic appearance defect detection. The AIGC method to generate defect data and the deep learning algorithm to realize automatic appearance defect detection are both technical means of the existing technology.

[0025] 360-degree surround imaging system: start the conveyor belt 1, which drives the can body 3 to move at a constant speed. When the can body 3 passes through the can separation sensor, the adjacent can bodies are separated to the required interval. When the can body 3 passes directly under the industrial camera 10 and multiple reflective lenses 5, photos are taken to collect pictures and record the conditions of the can body 3. The reflective lenses 5 are located at different angles of the can body and can record the appearance of the can body 3 at different angles. The inspected can body 3 can enter the rejection tank 8 under the push of the rejection cylinder 7.

[0026] At the same time, the data used by the main control center during detection is a self-built data set, and the defect images generated in advance using the AIGC method are used to simulate the defects that may occur in the actual production line, thereby greatly shortening the early data collection cycle and effectively solving the lag of the detection model. At the same time, different tank colors can be generated to improve the robustness of the trained model. The final model has the characteristics of small size, high accuracy and fast detection speed. In addition, in the process of building the model, it can adapt to the training of a small number of samples. Therefore, the model itself has the ability to update and iterate quickly.

[0027] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0028] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A tank appearance defect detection device based on deep learning and 360-degree surround imaging, characterized in that: The invention comprises a frame (6), wherein a detection component is detachably connected inside the frame (6), and a conveyor belt (1) and a can rejection receiving groove (8) are also connected inside the frame (6), wherein the conveyor belt (1) is located below the detection component, and a can separation mechanism is arranged at one end of the conveyor belt (1) close to the detection component, and the side of the conveyor belt (1) abuts against one end of the can rejection receiving groove (8) and the directions are perpendicular to each other and located on the same horizontal plane, and a can body rejection mechanism is fixedly installed on the side of the conveyor belt (1) away from the can rejection receiving groove (8), a can body (3) is arranged on the conveyor belt (1), and a main control center is also arranged on the frame (6), and the main control center is electrically connected to the detection component, and the detection component can detect defects of the can body (3), and the can separation mechanism can separate adjacent can bodies (3) to a suitable distance.

2. According to claim 1, a tank appearance defect detection device based on deep learning and 360-degree surround imaging is characterized in that: The detection component comprises an industrial camera (10), a fixing rod (4), a reflector fixing bracket and a reflective lens (5); a plurality of fixing rods (4) are fixedly arranged on the inner walls on both sides of the frame (6) close to the conveyor belt (1); the end of the fixing rod away from the frame (6) is connected to the reflector fixing bracket; the reflector fixing brackets are correspondingly distributed on both sides above the conveyor belt (1); the reflector fixing brackets are fixedly connected to the reflector lenses (5); and the industrial camera (10) is fixed to the top of the frame (6).

3. According to claim 1, a tank appearance defect detection device based on deep learning and 360-degree surround imaging is characterized in that: The can body rejection mechanism comprises a rejection cylinder connecting frame and a rejection cylinder (7); the rejection cylinder connecting frame is fixed to a side of the conveyor belt (1) away from the rejection can receiving groove (8); the rejection cylinder (7) is arranged on the rejection cylinder connecting frame; and the rejection cylinder (7) is electrically connected to a main control center.

4. The tank appearance defect detection device based on deep learning and 360-degree surround imaging according to claim 1 is characterized in that: The tank separation mechanism comprises a tank separation sensor, a tank separation push rod support frame and a tank separation push rod (9); the tank separation push rod support frame is fixed on the side wall of the conveyor belt (1); the tank separation push rod (9) and the tank separation sensor are connected to the tank separation push rod support frame; the tank separation push rod (9) is a pneumatic push rod; and the tank separation push rod (9) and the tank separation sensor are both electrically connected to a main control center.

5. The tank appearance defect detection device based on deep learning and 360-degree surround imaging according to claim 1 is characterized in that: An industrial light source is also provided inside the frame (6) above the conveyor belt (1), and the industrial light source is electrically connected to the main control center.

Images