Intelligent vision-controlled bottle product flaw detection and automatic rejection system
Patent Information
- Application Number
- CN202611120371.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-27
- Publication Date
- 2026-08-28
AI Technical Summary
[0005]鉴于此,本发明提出了一种智能视觉控制的瓶装产品瑕疵检测与自动剔除系统,旨在解决由于输送设备长期运行过程中存在电机转速波动、输送带张紧程度变化、设备振动以及产品间距变化等情况,瓶装产品的实际运动状态与理论输送状态之间往往存在差异,使得基于固定延时或固定距离触发的剔除方式容易出现剔除时机偏移的问题
[0032]Compared with existing technologies, the advantages of this invention are as follows: The product detection module divides the bottled product image into regions, acquiring image feature information for the bottle body, cap, and label areas respectively, and generating defect status information including defect type, location, and size. This not only determines whether the bottled product has defects but also identifies the specific state of the defects. The status marking module generates a dynamic status identifier for the corresponding defective bottled product after determining that it meets the rejection criteria. This associates the detection location, detection time, conveying status parameters, and defect status information, ensuring that the defect detection results correspond to the movement state during subsequent conveying. This enables continuous tracking of individual defective bottled products and reduces target identification errors during continuous multi-product conveying. The trajectory prediction module acquires continuous positional change information of the defective bottled product and the conveying equipment's operating status information. Based on the deviation between the actual movement state and the preset movement state, the predicted movement trajectory is dynamically corrected. This allows for adjustment of the predicted movement of the defective bottled product according to the actual operation of the conveying equipment, improving the accuracy of predicting the time it takes for defective products to reach the rejection area under high-speed conveying conditions. The rejection control module generates rejection control commands based on the predicted target time and achieves precise trigger control by combining this with the action process of the rejection actuator. This ensures that the rejection action matches the actual arrival time of defective bottled products in the rejection area, reducing missed or incorrect rejections caused by actuator response delays and improving the reliability of the automatic rejection process. By detecting the product status within the rejection area and providing feedback on the rejection results, rejection anomaly information is generated when defective bottled products are detected that have not been completely rejected. This allows for timely detection of rejection failures and improves the quality control capabilities of the production line.
Smart Images

Figure CN122644314A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of bottled product inspection technology, and more specifically, to an intelligent vision-controlled system for detecting and automatically rejecting defects in bottled products. Background Technology
[0002] With the development of automated production technology, the production and processing of bottled products in industries such as food, beverage, pharmaceuticals, and daily chemicals are gradually moving towards higher speeds and continuous processes. To ensure product quality, production lines typically need to inspect bottled products for appearance defects during the conveying process and automatically remove defective products from the conveyor line.
[0003] Existing patent CN119327757A discloses an intelligent production line rejection system based on image acquisition and analysis, including a conveying mechanism, a visual inspection mechanism, a control module, and a mechanical rejection mechanism. The visual inspection mechanism acquires images of bottled products in the inspection area and determines whether the products have defects according to preset inspection standards. When a defective product is detected, the control module controls the rejection mechanism to execute an action according to a preset delay time or a fixed trigger distance. However, in actual production, due to fluctuations in motor speed, changes in conveyor belt tension, equipment vibration, and changes in product spacing during long-term operation of the conveying equipment, the actual movement state of the bottled products often differs from the theoretical conveying state. This makes rejection methods based on fixed delays or fixed distance triggers prone to rejection timing deviations.
[0004] Therefore, it is necessary to design an intelligent vision-controlled system for detecting and automatically rejecting defects in bottled products to solve the problems existing in the current technology. Summary of the Invention
[0005] In view of this, the present invention proposes an intelligent vision-controlled system for detecting and automatically rejecting defects in bottled products. It aims to solve the problem that, due to fluctuations in motor speed, changes in conveyor belt tension, equipment vibration, and changes in product spacing during long-term operation of conveying equipment, there is often a difference between the actual motion state and the theoretical conveying state of bottled products. This makes rejection methods based on fixed delays or fixed distance triggers prone to rejection timing deviations.
[0006] This invention proposes an intelligent vision-controlled system for detecting and automatically rejecting defects in bottled products, comprising:
[0007] The product inspection module is used to detect trigger events of bottled products entering the inspection area, and to acquire images of the bottled products based on the trigger events; and to determine the defect status information of the bottled products based on the images of the bottled products.
[0008] The status marking module is used to obtain the detection location, detection time and transport status parameters of the bottled product when the defect status information meets the rejection conditions, to regard the bottled product that meets the rejection conditions as a defective bottled product, and to generate a dynamic status mark of the defective bottled product.
[0009] The trajectory prediction module is used to obtain continuous position change information of the defective bottled product and the current operating status information of the conveying equipment based on the dynamic status identifier; determine the actual movement state of the defective bottled product based on the continuous position change information, and determine whether there is a deviation between the actual movement state and the preset movement state based on the operating status information; when there is a deviation, correct the predicted movement trajectory of the defective bottled product based on the operating status information; and determine the target time for the defective bottled product to reach the rejection area based on the predicted movement trajectory.
[0010] The rejection control module is used to generate rejection control instructions based on the target time and control the rejection execution mechanism to perform rejection actions according to the rejection control instructions.
[0011] Furthermore, when the product inspection module determines the defect status information of the bottled product based on the image of the bottled product, it includes:
[0012] The product detection module divides the bottled product image into regions, namely the bottle body region, the bottle cap region, and the label region, and extracts image feature information of the bottle body region, the bottle cap region, and the label region. Based on the image feature information, it determines the defect status information of the bottled product. The defect status information includes defect type, defect location, and defect size.
[0013] Furthermore, when the status marking module determines whether the defect status information meets the rejection conditions, it includes:
[0014] The status marking module obtains a preset defect evaluation standard, which includes a defect type blacklist, a region sensitivity coefficient, and a basic size tolerance threshold.
[0015] When the defect type belongs to the fatal defect type in the defect type blacklist, the defect status information is determined to meet the removal condition; when the defect type does not belong to the fatal defect type, the area sensitivity coefficient of the area where the defect is located is determined according to the defect location, and the basic size tolerance threshold is corrected according to the area sensitivity coefficient to obtain the target tolerance threshold; when the defect size is greater than or equal to the target tolerance threshold, the defect status information is determined to meet the removal condition.
[0016] Furthermore, when the status marking module generates the dynamic status identifier for the defective bottled product, it includes:
[0017] The status marking module assigns a unique tracking code to the defective bottled product; converts the detection position into initial spatial coordinates, the detection time into an initial timestamp, and parses the delivery status parameters into an initial motion vector; binds the defect status information, the initial spatial coordinates, the initial timestamp, and the initial motion vector to construct a multi-dimensional status feature vector associated with the tracking code; and encapsulates the multi-dimensional status feature vector into the dynamic status identifier.
[0018] Furthermore, when the trajectory prediction module acquires the continuous position change information, it includes:
[0019] The trajectory prediction module performs position matching on the defective bottled product based on the tracking code in the dynamic status identifier; as the defective bottled product moves along the conveying direction, it obtains the position coordinate information of the defective bottled product at several consecutive moments based on the initial spatial coordinates and the initial timestamp, according to a preset sampling period; it determines the position change of the defective bottled product based on the position coordinate information and the initial spatial coordinates; and it generates the continuous position change information based on the position change within several consecutive sampling periods.
[0020] Furthermore, when the trajectory prediction module determines the actual movement state of the defective bottled product, it includes:
[0021] The trajectory prediction module determines the actual moving speed of the defective bottled product based on the position change between adjacent sampling times; determines the stable moving state of the defective bottled product based on the changes in the actual moving speed over several consecutive sampling periods; and determines the actual moving state of the defective bottled product based on the actual moving speed and the stable moving state.
[0022] Furthermore, when the trajectory prediction module determines whether there is a deviation between the actual movement state and the preset motion state, it includes:
[0023] The operating status information includes operating speed information and operating fluctuation information;
[0024] The trajectory prediction module determines the preset moving speed of the current conveying equipment based on the running speed information, and determines the motion state benchmark of the current conveying equipment based on the running fluctuation information; it compares the actual moving speed in the actual moving state with the preset moving speed, and matches the moving stable state with the motion state benchmark;
[0025] When the difference between the actual moving speed and the preset moving speed exceeds the preset speed deviation range, or when the moving stable state does not match the motion state benchmark, it is determined that there is a deviation between the actual moving state and the preset motion state.
[0026] Furthermore, when the trajectory prediction module corrects the predicted motion trajectory of the defective bottled product based on the running status information, it includes:
[0027] The trajectory prediction module determines the trajectory offset state of the defective bottled product based on the running speed information, the initial motion vector, and the actual moving speed; adjusts the predicted motion trajectory of the defective bottled product according to the trajectory offset state, and updates the expected position change process of the defective bottled product in the conveying direction; and generates a corrected predicted motion trajectory based on the adjusted expected position change process.
[0028] Furthermore, when the rejection control module generates a rejection control command based on the target time, it includes:
[0029] The rejection control module acquires the response time information of the rejection execution mechanism and determines the rejection trigger time based on the target time and the response time information. When the rejection trigger time is reached, the rejection control instruction is generated and sent to the rejection execution mechanism to control the rejection execution mechanism to perform rejection action on the defective bottled products.
[0030] Furthermore, it also includes:
[0031] The rejection control module collects product status information within the rejection area and determines whether the defective bottled products have been rejected. When it is detected that the defective bottled products have not been rejected, rejection exception information is generated.
[0032] Compared with existing technologies, the advantages of this invention are as follows: The product detection module divides the bottled product image into regions, acquiring image feature information for the bottle body, cap, and label areas respectively, and generating defect status information including defect type, location, and size. This not only determines whether the bottled product has defects but also identifies the specific state of the defects. The status marking module generates a dynamic status identifier for the corresponding defective bottled product after determining that it meets the rejection criteria. This associates the detection location, detection time, conveying status parameters, and defect status information, ensuring that the defect detection results correspond to the movement state during subsequent conveying. This enables continuous tracking of individual defective bottled products and reduces target identification errors during continuous multi-product conveying. The trajectory prediction module acquires continuous positional change information of the defective bottled product and the conveying equipment's operating status information. Based on the deviation between the actual movement state and the preset movement state, the predicted movement trajectory is dynamically corrected. This allows for adjustment of the predicted movement of the defective bottled product according to the actual operation of the conveying equipment, improving the accuracy of predicting the time it takes for defective products to reach the rejection area under high-speed conveying conditions. The rejection control module generates rejection control commands based on the predicted target time and achieves precise trigger control by combining this with the action process of the rejection actuator. This ensures that the rejection action matches the actual arrival time of defective bottled products in the rejection area, reducing missed or incorrect rejections caused by actuator response delays and improving the reliability of the automatic rejection process. By detecting the product status within the rejection area and providing feedback on the rejection results, rejection anomaly information is generated when defective bottled products are detected that have not been completely rejected. This allows for timely detection of rejection failures and improves the quality control capabilities of the production line. Attached Figure Description
[0033] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0034] Figure 1 A structural block diagram of an intelligent vision-controlled bottled product defect detection and automatic rejection system provided in an embodiment of the present invention;
[0035] Figure 2 A flowchart for the detection and automatic rejection of defects in bottled products provided in an embodiment of the present invention;
[0036] Figure 3 This is a schematic diagram of the structure of the bottled product defect detection and automatic rejection device provided in an embodiment of the present invention.
[0037] Among them, 1. Diffuse reflection photoelectric sensor; 2. Visual inspection sensor; 3. Cylinder-type rejection actuator; 4. Conveying equipment; 5. Waste collection area. Detailed Implementation
[0038] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0039] In some embodiments of this application, see Figure 1-2 As shown, an intelligent vision-controlled system for detecting and automatically rejecting defects in bottled products is proposed, comprising:
[0040] The product inspection module is used to detect trigger events when bottled products enter the inspection area, and to acquire images of the bottled products based on the trigger events; and to determine the defect status information of the bottled products based on the images of the bottled products.
[0041] The status marking module is used to obtain the detection location, detection time and conveying status parameters of the bottled product when the defect status information meets the rejection conditions, to identify the bottled product that meets the rejection conditions as a defective bottled product, and to generate a dynamic status mark of the defective bottled product.
[0042] The trajectory prediction module is used to obtain continuous position change information of defective bottled products and current operating status information of the conveying equipment based on dynamic status identifiers; determine the actual movement status of defective bottled products based on continuous position change information, and judge whether there is a deviation between the actual movement status and the preset movement status based on the operating status information; when there is a deviation, correct the predicted movement trajectory of defective bottled products based on the operating status information; and determine the target time for defective bottled products to reach the rejection area based on the predicted movement trajectory.
[0043] The rejection control module is used to generate rejection control instructions based on the target time and control the rejection execution mechanism to perform rejection actions according to the rejection control instructions.
[0044] Specifically, as the bottled product moves with the conveyor equipment, when it enters the preset detection area, the product detection module first detects the trigger event of the bottled product entering the detection area. This trigger event can be generated by a photoelectric detection element, a proximity detection element, or a position detection element located at the entrance of the detection area. When the detection element detects that the bottled product has entered the preset detection area, it sends a trigger signal to the product detection module. Based on the trigger signal, the product detection module controls the image acquisition unit to start, thereby acquiring image data of the corresponding bottled product.
[0045] To ensure that the acquired images of the bottled product reflect the appearance of different areas of the product, multiple image acquisition directions are set up in the detection area, and images of the bottled product are acquired through at least one image acquisition unit. The image acquisition unit may include an industrial camera, a lens, and a light source assembly. The light source assembly provides a stable lighting environment for the bottled product to reduce the impact of ambient light changes during transportation on image quality. The image data acquired by the image acquisition unit includes at least the outer surface contour information, color change information, texture information, and local defect area information of the bottled product.
[0046] After receiving the image of the bottled product, the product inspection module preprocesses the image to improve the accuracy of subsequent defect identification. Specifically, this includes image resizing, noise filtering, brightness correction, and effective region extraction. Resizing adjusts images acquired under different conditions to a uniform analysis size; noise filtering reduces invalid image information caused by equipment vibration and environmental interference during image acquisition; and brightness correction minimizes the impact of lighting changes on image feature extraction.
[0047] The product inspection module divides the pre-processed bottled product image into regions based on the structural characteristics of the bottled product, namely the bottle body region, the cap region, and the label region. The bottle body region characterizes the integrity of the bottle surface and any appearance defects; the cap region characterizes the cap's installation status, deformation, and sealing structure; and the label region characterizes the label's integrity, misalignment, and surface anomalies.
[0048] Specifically, when dividing the area, the product detection module determines the range of each detection area based on the pre-stored structural location information of the bottled product, or determines the boundary position between different areas based on the edge contour, color distribution and structural features in the bottled product image, thereby obtaining the corresponding bottle area image, cap area image and label area image.
[0049] The product inspection module extracts image features from the bottle body, cap, and label areas to obtain image feature information characterizing the state of each area. Specifically, for the bottle body, it extracts features such as surface texture changes, contour integrity, abnormal spot distribution, and surface damage. For the cap, it extracts features such as cap contour, edge shape, installation position, and color change. For the label, it extracts features such as label boundary position, label integrity, text area changes, and label pasting offset.
[0050] The product inspection module matches and analyzes the extracted image feature information with preset product standard state information. This product standard state information is established based on normal bottled product samples and describes the region feature range corresponding to a flawless state of the bottled product. When the image feature information corresponding to the detected area deviates from the product standard state information, an anomaly is determined in the corresponding area, and the defect state information of the bottled product is determined based on the feature changes corresponding to the anomaly area.
[0051] Specifically, the defect status information includes defect type, defect location, and defect size. Defect type indicates the category of anomaly present in the bottled product, such as bottle breakage, cap abnormality, missing label, misaligned label, or surface contamination. Defect location indicates the spatial position of the defect within the bottled product. Defect size indicates the size of the area corresponding to the defect, allowing for subsequent determination of whether the bottled product needs to be rejected based on the severity of the defect.
[0052] Specifically, the status marking module is used to determine whether the corresponding bottled product meets the rejection criteria based on the defect status information output by the product detection module. When it is determined that the product needs to be rejected, a dynamic status mark is established for the bottled product so that the subsequent trajectory prediction module can continuously track the defective bottled product.
[0053] Specifically, after receiving the defect status information sent by the product inspection module, the status marking module parses the defect status information to obtain the defect type, defect location, and defect size information contained therein, and then calls the pre-stored defect evaluation criteria to judge the defect status information. The defect evaluation criteria describe the rejection rules corresponding to different types of defects, different product areas, and different defect sizes.
[0054] The defect assessment criteria include a defect type blacklist, regional sensitivity information, and basic size tolerance thresholds. The defect type blacklist stores unacceptable defect types that affect the safety or functionality of bottled products, such as bottle breakage, damaged bottle necks, and missing caps. When a detected defect type matches any type in the defect type blacklist, the status marking module directly determines the corresponding bottled product as a defective product and generates a corresponding rejection flag.
[0055] When the detected defect type does not belong to the fatal defect type in the defect type blacklist, the status marking module further determines the detection area where the defect is located based on the defect location and obtains the corresponding area sensitivity coefficient. The area sensitivity coefficient represents the tolerance of different areas of the bottled product to changes in defect size. Different areas correspond to different sensitivity levels, determined based on the importance of different areas of the bottled product and historical quality inspection data. For example, for areas affecting the product's sealing performance, such as the bottle neck sealing area and the cap connection area, the area sensitivity coefficient is set higher to improve the sensitivity of this area to small-sized defects; for non-critical appearance areas of the bottle body, the area sensitivity coefficient is set lower to avoid the wrong rejection of normal products due to slight appearance changes.
[0056] Specifically, when the status marking module adjusts the basic size tolerance threshold based on the region sensitivity coefficient, it first obtains the basic size tolerance threshold corresponding to the current defect type. The basic size tolerance threshold represents the maximum allowable defect size range without considering regional differences. Subsequently, it determines the threshold adjustment direction based on the region sensitivity coefficient corresponding to the region where the defect is located.
[0057] When the area sensitivity coefficient is higher than the preset sensitivity level, it indicates that the area is more sensitive to defects. The status marking module lowers the basic size tolerance threshold so that smaller defects can meet the rejection criteria. When the area sensitivity coefficient is lower than the preset sensitivity level, it indicates that the area has less impact on product performance. The status marking module raises the basic size tolerance threshold so that minor appearance defects will not be misjudged as products that need to be rejected.
[0058] When adjusting the threshold, the status marking module determines the adjustment range based on the difference between the area sensitivity coefficient and the preset baseline sensitivity level. When the difference between the area sensitivity coefficient and the baseline sensitivity level is large, the basic size tolerance threshold is adjusted significantly; when the difference is small, the basic size tolerance threshold is adjusted slightly to obtain a target tolerance threshold that matches the current defective area. For example, in the same bottled product, the bottle opening sealing area corresponds to a higher area sensitivity coefficient, and the status marking module adjusts the basic size tolerance threshold to a smaller range to ensure that minor defects in the bottle opening area can be promptly rejected. For the bottle label edge area, since it mainly affects appearance and not product function, the status marking module adjusts the basic size tolerance threshold to a larger range to reduce the probability of normal products being mistakenly rejected.
[0059] When the defect size is greater than or equal to the target tolerance threshold after area correction, the status marking module determines that the current defect status information meets the rejection conditions and identifies the corresponding bottled product as a defective bottled product; when the defect size is less than the target tolerance threshold, the current defect status information does not meet the rejection conditions.
[0060] After determining that a bottled product meets the rejection criteria, the status marking module generates a dynamic status identifier for the defective bottled product. Specifically, the status marking module first assigns a unique tracking code to the defective bottled product. The tracking code is used to continuously associate the same defective bottled product during transportation, avoiding status confusion when multiple bottled products pass through the detection area at the same time.
[0061] The status marker module acquires the detection position, detection time, and conveying status parameters of the defective bottled product at the time of detection. The detection position indicates the spatial location of the bottled product on the conveying equipment when it is detected; the detection time indicates the time at which the defect is determined; and the conveying status parameters indicate the operating status of the conveying equipment at the time of detection, including conveying speed, direction of travel, and conveying stability.
[0062] The status marking module converts the detection position into spatial coordinate information to establish the initial spatial reference position of the defective bottled product within the detection area; it converts the detection time into corresponding timestamp information to determine the time base for subsequent changes in motion status; at the same time, it analyzes the transport status parameters, extracts motion information that can reflect the initial motion trend of the product, and generates the initial motion vector.
[0063] The status marking module associates and binds defect status information, initial spatial coordinates, initial timestamp, and initial motion vector with the tracking code to form multi-dimensional status information describing the detection status and initial motion status of the defective bottled product. This multi-dimensional status information includes at least product identity information, defect information, spatial location information, time information, and motion status information. Finally, the status marking module encapsulates the multi-dimensional status information to generate a dynamic status identifier corresponding to the defective bottled product and sends this identifier to the trajectory prediction module. This allows the trajectory prediction module to continuously acquire the position change information of the defective bottled product during transportation based on the same tracking code, achieving continuous status tracking of the defective bottled product.
[0064] Specifically, the trajectory prediction module is used to continuously track the motion of defective bottled products that meet the rejection criteria, and dynamically predict the subsequent motion trajectory of the defective bottled products based on changes in the operating status of the conveying equipment, thereby determining the target time for the defective bottled products to reach the rejection area, and providing an accurate execution sequence for the subsequent rejection control module.
[0065] Specifically, after receiving the dynamic status identifier sent by the status marker module, the trajectory prediction module parses the dynamic status identifier to obtain the tracking code, initial spatial coordinates, initial timestamp, and initial motion vector corresponding to the defective bottled product. The tracking code is used to uniquely identify the target defective bottled product when multiple bottled products continuously pass through the detection area, avoiding state confusion between different bottled products; the initial spatial coordinates represent the initial position of the defective bottled product on the conveyor equipment when it is identified; the initial timestamp represents the time reference for determining the defective state; and the initial motion vector represents the initial motion direction and trend of the defective bottled product at the detection moment.
[0066] The trajectory prediction module matches the location of defective bottled products based on tracking codes. In this embodiment, a tracking detection unit installed on the side of the conveying equipment continuously collects product status information within the conveying area. Based on the product's outer contour features, position features, or code association information, the subsequently collected product position data is matched with the corresponding tracking code to determine the continuous positional changes of the same defective bottled product during the conveying process.
[0067] As the defective bottled products move along the conveyor direction, the trajectory prediction module continuously collects data on the position of the defective bottled products according to a preset sampling period. The preset sampling period is determined based on the operating speed of the conveyor equipment and the interval between the rejection area and the detection area, ensuring that the collected positional change information meets the accuracy requirements for subsequent trajectory prediction.
[0068] Specifically, the trajectory prediction module uses the initial spatial coordinates in the dynamic status identifier as the reference benchmark for position change and the initial timestamp as the reference benchmark for time change, acquiring the position coordinate information corresponding to the defective bottled products at multiple consecutive sampling moments. The position coordinate information includes at least the position coordinates of the defective bottled products in the conveying direction, as well as auxiliary position coordinates used to indicate the lateral offset of the products.
[0069] The trajectory prediction module determines the positional change of the defective bottled product based on the positional coordinate information obtained at consecutive sampling times. Specifically, by comparing the positional coordinates corresponding to adjacent sampling times, it determines the positional change of the defective bottled product within a unit sampling period, and correlates the positional changes within multiple consecutive sampling periods to form continuous positional change information reflecting the movement process of the defective bottled product.
[0070] The trajectory prediction module determines the actual movement state of the defective bottled products based on continuous position change information. Specifically, the trajectory prediction module determines the actual moving speed of the defective bottled products within the current sampling period based on the position change between adjacent sampling times and the corresponding time interval. The actual moving speed reflects the actual speed of the defective bottled products during the conveying process, distinguishing it from the set operating speed of the conveying equipment.
[0071] The trajectory prediction module analyzes the actual moving speed acquired over multiple consecutive sampling periods and determines the stable movement state of the defective bottled product based on the speed changes between adjacent sampling periods. Specifically, when the actual moving speed remains within a preset range over multiple consecutive sampling periods, the defective bottled product is determined to be in a stable movement state; when the actual moving speed exhibits continuous changes or periodic fluctuations, the defective bottled product is determined to be in a fluctuating movement state.
[0072] The trajectory prediction module determines the actual movement state of the defective bottled product based on its actual moving speed and stability. The actual movement state comprehensively describes the current speed and stability of the defective bottled product, allowing for subsequent comparison with the preset movement state of the conveying equipment.
[0073] The trajectory prediction module determines whether there is a deviation between the actual movement state and the preset motion state based on the operating status information of the conveying equipment. Specifically, the trajectory prediction module acquires the operating status information of the conveying equipment, which includes operating speed information and operating fluctuation information. The operating speed information is used to indicate the current set or actual operating speed of the conveying equipment, and the operating fluctuation information is used to indicate the changes in operating stability of the conveying equipment during operation due to motor fluctuations, mechanical vibrations, or load changes.
[0074] The trajectory prediction module determines the preset moving speed of the conveying equipment based on the operating speed information, and determines the corresponding motion state benchmark based on the operating fluctuation information. The motion state benchmark describes the theoretically stable motion range that the bottled products should maintain under the current operating conditions of the conveying equipment.
[0075] The trajectory prediction module compares the actual movement speed with the preset movement speed during the actual movement state, and simultaneously matches the stable movement state with the motion state benchmark. When the difference between the actual movement speed and the preset movement speed is within the allowable range, and the stable movement state matches the motion state benchmark, it is determined that the defective bottled product moves according to the expected movement state; when the difference between the actual movement speed and the preset movement speed exceeds the preset speed deviation range, or the stable movement state does not match the motion state benchmark, it is determined that there is a deviation between the actual movement state and the preset motion state of the defective bottled product.
[0076] The motion state benchmark includes at least a benchmark moving speed range, a speed variation fluctuation range, and a continuous stable holding time. The benchmark moving speed range is determined based on the set operating speed of the conveying equipment and the equipment's allowable operating error; the speed variation fluctuation range is obtained by statistically analyzing historical speed data continuously collected by the conveying equipment under normal operating conditions, and is used to characterize the allowable speed variation amplitude during normal operation; the continuous stable holding time is used to determine whether the bottled product maintains a stable conveying state.
[0077] Specifically, the trajectory prediction module acquires the actual moving speed over multiple consecutive sampling periods and calculates the speed change between adjacent sampling periods. When the actual moving speed is within the baseline moving speed range, and the speed change within multiple consecutive sampling periods does not exceed the speed fluctuation range, while the stable holding time reaches a preset time threshold, the current stable moving state is determined to match the motion state baseline. Conversely, when the actual moving speed exceeds the baseline moving speed range, or the speed change continuously exceeds the speed fluctuation range, or the stable holding time requirement is not met, the stable moving state is determined to be mismatched with the motion state baseline. For example, when the conveyor is set to operate at 0.8 m / s, the baseline moving speed range is determined to be 0.8 m / s ± 0.03 m / s based on historical stable operating data, the speed fluctuation range is defined as a speed change not exceeding 0.02 m / s between adjacent sampling periods, and the continuous stable holding time is set to 500 ms. When the defective bottled product meets the above conditions within 500 ms, it is determined to move according to the preset motion state.
[0078] When no deviation is detected between the actual movement state and the preset movement state, there is no need to correct the predicted movement trajectory. The target time for the defective bottled products to reach the rejection area can be determined directly based on the predicted movement trajectory.
[0079] When a deviation is detected between the actual movement state and the preset movement state, the trajectory prediction module corrects the predicted movement trajectory of the defective bottled product based on the operating status information. Specifically, the trajectory prediction module acquires the operating speed information of the conveyor equipment, the initial motion vector in the dynamic status identifier, and the currently collected actual movement speed, and determines the trajectory offset state of the defective bottled product relative to the initial predicted path based on the above information.
[0080] The trajectory offset status is used to indicate the positional deviation of the defective bottled product from its theoretical movement process, including positional advance, positional lag, and changes in movement stability along the conveying direction.
[0081] The trajectory prediction module adjusts the original predicted motion trajectory based on the trajectory offset status. When the detected speed of the defective bottled product is higher than the preset speed, the subsequent predicted position change process is corrected in advance; when the detected speed of the defective bottled product is lower than the preset speed, the subsequent predicted position change process is corrected in a delayed manner; when the detected fluctuation in the operation of the conveyor equipment causes the product's motion state to be unstable, the subsequent predicted position is adjusted according to the position change trend obtained by continuous sampling.
[0082] The trajectory prediction module regenerates the corrected predicted motion trajectory based on the adjusted expected position change process, and determines the target time for defective bottled products to reach the rejection area based on the corrected predicted motion trajectory.
[0083] Specifically, the trajectory prediction module determines the positional relationship between the defective bottled product and the rejection area based on the current position of the defective bottled product, the corrected direction of movement, and the subsequent positional change trend. It also predicts the time node when the defective bottled product arrives at the rejection area and sends this time node as the target time to the rejection control module.
[0084] Specifically, when the trajectory prediction module adjusts the original predicted motion trajectory based on the trajectory offset status, it first obtains the actual position of the defective bottled product at the current sampling time and the predicted position at the corresponding sampling time in the original predicted motion trajectory, and determines the trajectory offset based on the distance difference between the actual position and the predicted position; and corrects the subsequent expected position change process based on the trajectory offset.
[0085] When the actual moving speed of the defective bottled product is higher than the preset moving speed, it indicates that the actual moving distance of the defective bottled product is greater than the original prediction result. The trajectory prediction module determines that the defective bottled product has an early offset along the conveying direction and adds the trajectory offset to the expected position corresponding to each subsequent prediction time, so that the subsequent expected position moves ahead along the conveying direction.
[0086] When the actual moving speed of the defective bottled product is lower than the preset moving speed, it indicates that the actual moving distance of the defective bottled product is less than the original prediction result. The trajectory prediction module determines that the defective bottled product has a lag offset in the opposite direction of the conveying and deducts the trajectory offset from the expected position corresponding to each subsequent prediction time, so that the subsequent expected position moves later in the conveying direction.
[0087] When fluctuations in the operation of the conveying equipment are detected, causing instability in the product's motion, the trajectory prediction module acquires the position change over multiple consecutive sampling periods and adjusts the distance between subsequent predicted positions based on the position change trend over consecutive sampling periods. When the position change over consecutive sampling periods continues to increase, the interval between subsequent predicted positions is increased; when the position change over consecutive sampling periods continues to decrease, the interval between subsequent predicted positions is decreased.
[0088] The trajectory prediction module regenerates the corrected predicted motion trajectory based on the adjusted expected position, and redetermines the target time for the defective bottled products to reach the rejection area based on the corrected predicted motion trajectory.
[0089] For example, the conveyor's preset moving speed is 0.8 m / s, and the trajectory prediction module uses a sampling period of 50 ms to generate the original predicted motion trajectory based on the initial detection position. Assuming the initial position of the defective bottled product is 1000 mm, under normal operating conditions of the conveyor, and based on the preset moving speed, the predicted positions at 50 ms, 100 ms, and 150 ms are 1040 mm, 1080 mm, and 1120 mm, respectively.
[0090] After 50ms of sampling, the trajectory prediction module obtains the actual position of the defective bottled product as 1048mm. Based on the initial position of 1000mm, the actual movement distance within the current sampling period is calculated to be 48mm, corresponding to an actual movement speed of 0.96m / s, which is higher than the preset movement speed of 0.8m / s. Therefore, it is determined that the defective bottled product has shifted prematurely relative to the original predicted motion trajectory. Further, based on the difference between the actual position of 1048mm and the predicted position of 1040mm, the trajectory offset is determined to be 8mm, and this trajectory offset is added as a correction to the subsequent predicted positions. While maintaining the current trend in the conveying state, the predicted positions corresponding to 100ms and 150ms are adjusted from 1080mm and 1120mm to 1088mm and 1128mm, respectively. The predicted motion trajectory is then regenerated based on the adjusted predicted positions.
[0091] When the actual position of the defective bottled product was detected as 1035mm during subsequent sampling, while the corresponding original predicted position was 1040mm, the actual movement distance calculated based on the initial position of 1000mm was 35mm, corresponding to an actual movement speed of 0.7m / s, which is lower than the preset movement speed of 0.8m / s. Therefore, it was determined that the defective bottled product had a lag offset relative to the original predicted motion trajectory. At this time, the trajectory offset was determined to be 5mm based on the difference between the actual position of 1035mm and the predicted position of 1040mm. This trajectory offset was then subtracted from the subsequent predicted positions as a delay correction. The predicted positions corresponding to 100ms and 150ms were adjusted from 1080mm and 1120mm to 1075mm and 1115mm, respectively, to obtain the corrected predicted motion trajectory.
[0092] When fluctuations in the operation of the conveyor equipment are detected, causing instability in the product's motion, the trajectory prediction module acquires the position change within a continuous sampling period and adjusts the subsequent predicted positions based on the position change trend. For example, if the position changes of the defective bottled product are 40mm, 42mm, 45mm, and 48mm over four consecutive sampling periods, it indicates that the conveyor equipment is gradually accelerating. Based on this trend, the trajectory prediction module adjusts the subsequent predicted position intervals from a fixed 40mm to 42mm, 45mm, and 48mm. If the subsequent three predicted positions in the original predicted trajectory are 1120mm, 1160mm, and 1200mm, the corrected predicted positions are recalculated based on the adjusted position changes to 1122mm, 1167mm, and 1215mm, and a new predicted motion trajectory is generated based on these corrected predicted positions.
[0093] Finally, the trajectory prediction module re-determines the target time for the defective bottled product to reach the rejection area based on the corrected predicted motion trajectory. For example, before correction, the remaining distance between the defective bottled product and the rejection area was 2000mm, and the estimated arrival time was 2.50s based on the preset moving speed of 0.8m / s. After trajectory correction, the actual remaining moving distance of the defective bottled product is determined to be 2010mm based on the corrected trajectory. At the same time, based on the continuous position change information, the subsequent predicted moving speed is adjusted to 0.83m / s. Therefore, the recalculated target time is 2010mm ÷ 830mm / s ≈ 2.42s, and 2.42s is sent to the rejection control module as the new target time to control the rejection execution mechanism to perform the rejection action at the corresponding time.
[0094] Specifically, the rejection control module controls the rejection execution mechanism to perform actions based on the target time determined by the trajectory prediction module, ensuring that defective bottled products are accurately separated when they reach the rejection area. Compared to existing technologies that trigger the rejection mechanism based on a fixed distance or a fixed delay time, this embodiment dynamically determines the rejection execution sequence based on the actual movement trajectory of the defective bottled products, thereby reducing rejection deviations caused by factors such as conveyor speed fluctuations and equipment response delays.
[0095] Specifically, the rejection control module receives target time information from the trajectory prediction module. The target time represents the time required for the defective bottled product to move to the rejection area according to the corrected predicted trajectory. Based on the target time, the rejection control module determines the trigger moment for controlling the rejection actuator, ensuring that the actuator completes its action when the defective bottled product enters the effective rejection range.
[0096] Since the rejection actuator does not immediately perform an actual rejection action after receiving the control signal, but rather experiences response delays such as signal transmission delay, controller processing delay, solenoid valve response delay, and mechanical actuator movement delay, the rejection control module pre-obtains the response time information corresponding to the rejection actuator. This response time information represents the time interval between the output of the rejection control command and the rejection actuator reaching a valid rejection state.
[0097] Specifically, response time information can be obtained through the factory parameters of the rejection actuator or through statistics from historical execution processes. In one implementation, the rejection control module records the time difference between the control command sending time and the actual completion time of the rejection actuator during multiple rejection operations, and determines the response time information corresponding to the current rejection actuator based on multiple historical response time data.
[0098] The rejection control module determines the rejection trigger time based on the target time and response time information. Specifically, the rejection control module uses the target time when the defective bottled product arrives at the rejection area as the action completion benchmark, and combines this with the response time required for the rejection actuator to complete the action, advancing the control command sending time to the time node when the rejection actuator can complete the rejection action within the target time. For example, when the trajectory prediction module determines that the defective bottled product will enter the rejection area at a certain time, and the rejection actuator needs a certain amount of time from receiving the control command to completing the action, the rejection control module generates the rejection trigger time in advance, so that the control signal can act on the rejection actuator in advance, thereby offsetting the actuator's own response delay.
[0099] When the system runtime reaches the rejection trigger time, the rejection control module generates a corresponding rejection control command and sends it to the rejection execution mechanism. The rejection control command includes at least action trigger information and execution object information, instructing the rejection execution mechanism to perform rejection operations on defective bottled products with corresponding tracking codes.
[0100] Specifically, the rejection actuator may include a cylinder, solenoid valve, mechanical lever, pusher mechanism, or other actuators capable of altering the conveying path of bottled products. Upon receiving a rejection control command, the rejection actuator operates according to a preset action mode, causing the defective bottled products to deviate from the normal conveying path and enter the waste collection area.
[0101] To improve the reliability of the rejection process, a rejection status feedback detection process is also set up. After sending the rejection control command, the rejection control module continues to collect product status information within the rejection area and determines whether the corresponding defective bottled products have been rejected based on the product status information.
[0102] Specifically, a status detection unit is set up in the rejection area to acquire information on changes in the position of products within the rejection area. This status detection unit can employ photoelectric detection elements, visual detection elements, or position detection elements. When a defective bottled product is detected leaving its original transport path and entering a preset collection area, it is determined that the defective bottled product has been rejected. When a defective bottled product with the corresponding tracking code is detected still in the normal transport path, or is not detected entering the preset collection area, it is determined that the defective bottled product has not been rejected.
[0103] When defective bottled products are detected and not completely rejected, the rejection control module generates rejection anomaly information and records the corresponding anomaly status. The rejection control module can send the rejection anomaly information to the control system to prompt equipment maintenance personnel to conduct an inspection.
[0104] In one specific embodiment, see Figure 3 As shown, this system is used for detecting appearance defects in bottled products and for automatic sorting and rejection in beverage production lines. The system includes a product detection module, a status marking module, a trajectory prediction module, and a rejection control module. The system hardware includes a Keyence IV4 smart camera controller, a visual inspection sensor 2, a diffuse reflection photoelectric sensor 1, a solenoid valve, and a cylinder-type rejection actuator 3.
[0105] Bottled products move continuously via conveyor 4, which operates at a speed of 0.8 m / s, maintaining an interval of approximately 150 mm between each bottle. An IFM O6H301 diffuse reflection photoelectric sensor 1 is installed at the entrance of the detection area on conveyor 4 to detect whether a bottled product has entered the visual detection area. When the photoelectric sensor detects a bottled product entering the detection area, it outputs an external trigger signal to the Keyence IV4-G120 intelligent camera controller.
[0106] After receiving an external trigger signal, the Keyence IV4-G120 smart camera controller controls the visual inspection sensor 2 to acquire images according to preset trigger logic. In this embodiment, the appropriate model of inspection sensor is selected based on the installation distance between the bottled product and the inspection sensor. When the detection distance is short, the IV4-G500CA inspection sensor is used; when the detection distance is long, the IV4-G600CA inspection sensor is used. In this embodiment, the installation distance of the inspection sensor is set to 300mm, and the IV4-G500CA is used to complete the image acquisition of the bottled product.
[0107] After acquiring images of bottled products, the visual inspection sensor 2 sends the acquired image data to the IV4-G120 smart camera controller, and processes and analyzes the images using Keyence IVSmartNavigator software. Specifically, the product inspection module first divides the acquired bottled product images into regions, namely the bottle body region, the cap region, and the label region, based on the structural characteristics of the bottled product.
[0108] The bottle body area is mainly used to detect defects such as bottle breakage, dents, contamination, and abnormal color; the bottle cap area is mainly used to detect defects such as missing bottle caps, tilted bottle caps, and abnormal bottle mouth seals; and the label area is mainly used to detect defects such as missing labels, offset labels, damaged labels, and abnormal printing.
[0109] The product detection module extracts image feature information from the aforementioned areas. For the bottle area, it extracts bottle edge integrity features, surface texture change features, and abnormal area size information; for the bottle cap area, it extracts bottle cap contour features, bottle cap position features, and installation status features; and for the label area, it extracts label edge position, label coverage area, and text area integrity features.
[0110] For example, when a damaged area is detected in the label area of a bottled product, the product detection module obtains the location coordinates of the defective area as a range of 30mm × 15mm to the right of the label area, and determines the defect type as "label damage" and the defect size as 450mm², thereby generating defect status information that includes defect type, defect location and defect size.
[0111] After receiving the aforementioned defect status information, the status marking module performs a rejection condition judgment on the defect status information. In this embodiment, a defect evaluation standard is established in advance, which includes a defect type blacklist, including fatal defect types that affect product safety, such as bottle breakage, missing bottle caps, and damaged bottle mouths.
[0112] When the detected defect type belongs to the aforementioned critical defect types, the status marking module directly determines that the bottled product meets the rejection criteria. For example, when a missing bottle cap is detected, there is no need to further determine the defect size; a rejection mark is generated directly.
[0113] When the detected defect type is not a fatal defect type, the status marking module further determines the corresponding area sensitivity coefficient based on the defect location. For example, the area sensitivity coefficient is set to 1.5 for the bottle mouth sealing area, 0.8 for the label area, and 0.6 for the non-critical area of the bottle body.
[0114] The status marking module adjusts the basic size tolerance threshold based on the area sensitivity coefficient. In this embodiment, the basic size tolerance threshold is set to 500 mm² for the label area. When a defect is detected in the label area, since the label area sensitivity coefficient is 0.8, indicating that the area has a low impact on functionality, the basic size tolerance threshold is adjusted to 600 mm². When a defect is detected in the bottle mouth sealing area, since the area sensitivity coefficient is 1.5, the basic size tolerance threshold is adjusted to 300 mm².
[0115] When the detected defect size is greater than or equal to the adjusted target tolerance threshold, the status marking module determines that the bottled product meets the rejection criteria and marks the bottled product as a defective bottled product.
[0116] The status marking module generates dynamic status identifiers for bottled products that need to be rejected. For example, it assigns a tracking code ID-000125 to the current defective bottled product, obtains the position coordinates of the bottled product at the visual inspection location as X=1200mm, Y=0mm, and records the inspection timestamp T=15:30:20.250s.
[0117] The status marker module acquires the current operating status parameters of the conveyor device 4, including a conveying speed of 0.8 m / s at the detection time, a forward conveying direction, and a device operating fluctuation range of ±0.03 m / s. The status marker module binds the above detection location, detection time, conveying speed, and defect status information to form a dynamic status identifier corresponding to the tracking code ID-000125.
[0118] Furthermore, the trajectory prediction module continuously tracks the defective bottled product based on the tracking code in the dynamic status identifier. In this embodiment, the trajectory prediction module acquires continuous position change information of the bottled product at a sampling period of 50ms. For example, the position coordinates acquired in five consecutive sampling periods are as follows: Sampling time 1: X=1240mm; Sampling time 2: X=1280mm; Sampling time 3: X=1322mm; Sampling time 4: X=1365mm; Sampling time 5: X=1405mm.
[0119] The trajectory prediction module determines the actual moving speed based on the position change between adjacent sampling times. For example, if the position change between the first and second sampling periods is 40 mm, the corresponding actual moving speed is 0.8 m / s.
[0120] The trajectory prediction module determines the stable movement state based on the speed changes over multiple consecutive sampling periods. For example, if the actual moving speeds are 0.80 m / s, 0.81 m / s, 0.79 m / s, and 0.80 m / s over 10 consecutive sampling periods, the defective bottled product is determined to be in a stable movement state.
[0121] The trajectory prediction module acquires the current operating status information of the conveyor 4, including an operating speed of 0.8 m / s and an allowable fluctuation range of ±0.03 m / s. The trajectory prediction module compares the actual moving speed with the preset moving speed.
[0122] When the actual moving speed is 0.95m / s, the difference between it and the preset moving speed of 0.8m / s is 0.15m / s, which exceeds the preset speed deviation range. Therefore, it is determined that there is a deviation between the actual moving state and the preset motion state.
[0123] The trajectory prediction module corrects the predicted motion trajectory based on the operating status information. For example, if the defective bottled product moves 50mm earlier than the initial predicted position due to the increased conveying speed, the trajectory prediction module adjusts the subsequent position change process based on the actual moving speed and the initial motion vector, correcting the expected time to reach the rejection area from the original prediction of 2.5s to 2.3s.
[0124] The trajectory prediction module sends the corrected predicted motion trajectory to the rejection control module and determines the target time for the defective bottled product to arrive at the rejection area. The rejection control module obtains the response time information of the rejection actuator. In this embodiment, the response time of the rejection actuator, which consists of a cylinder, a solenoid valve, and control signal transmission, has been determined to be 80ms after multiple tests.
[0125] When the trajectory prediction module determines that the defective bottled product will arrive at the rejection area at 15:30:22.550s, the rejection control module calculates the rejection trigger time as 15:30:22.470s based on the 80ms response time. When the system time reaches 15:30:22.470s, the rejection control module outputs a rejection control command to the solenoid valve. The solenoid valve controls the cylinder to extend, causing the cylinder push rod to act on the defective bottled product, pushing it from the normal conveying path into the waste collection area 5.
[0126] To ensure the reliable completion of the rejection process, the rejection control module continues to collect product status information in the rejection area after the rejection action is executed. In this embodiment, a detection sensor is set up in the rejection area. When the bottled product corresponding to ID-000125 is detected entering the waste collection area 5, the rejection is determined to be complete. If the bottled product is still detected in the normal conveying path within the preset detection time of 300ms, the rejection is determined to have failed, and rejection anomaly information is generated.
[0127] In summary, the product inspection module divides the bottled product image into regions, acquiring image feature information for the bottle body, cap, and label areas, and generating defect status information including defect type, location, and size. This not only determines whether the bottled product has defects but also identifies the specific state of the defects. The status marking module generates dynamic status markers for the corresponding defective bottled products after determining that they meet the rejection criteria. It correlates the detection location, detection time, conveying status parameters, and defect status information, ensuring that the defect detection results correspond to the movement state during subsequent conveying. This enables continuous tracking of individual defective bottled products and reduces target identification errors during continuous multi-product conveying. The trajectory prediction module acquires continuous positional change information of defective bottled products and conveying equipment operating status information. It dynamically corrects the predicted trajectory based on the deviation between the actual movement state and the preset movement state. This adjusts the predicted movement of defective bottled products according to the actual operation of the conveying equipment, improving the accuracy of predicting the arrival time of defective products in the rejection area under high-speed conveying conditions. The rejection control module generates rejection control commands based on the predicted target time and achieves precise trigger control by combining this with the action process of the rejection actuator. This ensures that the rejection action matches the actual arrival time of defective bottled products in the rejection area, reducing missed or incorrect rejections caused by actuator response delays and improving the reliability of the automatic rejection process. By detecting the product status within the rejection area and providing feedback on the rejection results, rejection anomaly information is generated when defective bottled products are detected that have not been completely rejected. This allows for timely detection of rejection failures and improves the quality control capabilities of the production line.
[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.
Claims
1. A smart vision-controlled system for detecting and automatically rejecting defects in bottled products, characterized in that, include: The product detection module is used to detect trigger events of bottled products entering the detection area and to acquire images of the bottled products based on the trigger events. Determine the defect status information of the bottled product based on the image of the bottled product; The status marking module is used to obtain the detection location, detection time and transport status parameters of the bottled product when the defect status information meets the rejection conditions, to regard the bottled product that meets the rejection conditions as a defective bottled product, and to generate a dynamic status mark of the defective bottled product. The trajectory prediction module is used to obtain continuous position change information of the defective bottled product and the current operating status information of the conveying equipment based on the dynamic status identifier; determine the actual movement state of the defective bottled product based on the continuous position change information, and determine whether there is a deviation between the actual movement state and the preset movement state based on the operating status information; when there is a deviation, correct the predicted movement trajectory of the defective bottled product based on the operating status information; and determine the target time for the defective bottled product to reach the rejection area based on the predicted movement trajectory. The rejection control module is used to generate rejection control instructions based on the target time and control the rejection execution mechanism to perform rejection actions according to the rejection control instructions.
2. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 1, characterized in that, When the product inspection module determines the defect status information of the bottled product based on the image of the bottled product, it includes: The product detection module divides the bottled product image into regions, namely the bottle body region, the bottle cap region, and the label region, and extracts image feature information of the bottle body region, the bottle cap region, and the label region. Based on the image feature information, it determines the defect status information of the bottled product. The defect status information includes defect type, defect location, and defect size.
3. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 2, characterized in that, When the status marking module determines whether the defect status information meets the rejection conditions, it includes: The status marking module obtains a preset defect evaluation standard, which includes a defect type blacklist, a region sensitivity coefficient, and a basic size tolerance threshold. When the defect type belongs to the fatal defect type in the defect type blacklist, the defect status information is determined to meet the removal condition; when the defect type does not belong to the fatal defect type, the area sensitivity coefficient of the area where the defect is located is determined according to the defect location, and the basic size tolerance threshold is corrected according to the area sensitivity coefficient to obtain the target tolerance threshold; when the defect size is greater than or equal to the target tolerance threshold, the defect status information is determined to meet the removal condition.
4. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 3, characterized in that, When the status marking module generates the dynamic status identifier for the defective bottled product, it includes: The status marking module assigns a unique tracking code to the defective bottled product; converts the detection position into initial spatial coordinates, the detection time into an initial timestamp, and parses the delivery status parameters into an initial motion vector; binds the defect status information, the initial spatial coordinates, the initial timestamp, and the initial motion vector to construct a multi-dimensional status feature vector associated with the tracking code; and encapsulates the multi-dimensional status feature vector into the dynamic status identifier.
5. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 4, characterized in that, When the trajectory prediction module acquires the continuous position change information, it includes: The trajectory prediction module performs position matching on the defective bottled product based on the tracking code in the dynamic status identifier; as the defective bottled product moves along the conveying direction, it obtains the position coordinate information of the defective bottled product at several consecutive moments based on the initial spatial coordinates and the initial timestamp, according to a preset sampling period; it determines the position change of the defective bottled product based on the position coordinate information and the initial spatial coordinates; and it generates the continuous position change information based on the position change within several consecutive sampling periods.
6. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 5, characterized in that, When the trajectory prediction module determines the actual movement state of the defective bottled product, it includes: The trajectory prediction module determines the actual moving speed of the defective bottled product based on the position change between adjacent sampling times; determines the stable moving state of the defective bottled product based on the changes in the actual moving speed over several consecutive sampling periods; and determines the actual moving state of the defective bottled product based on the actual moving speed and the stable moving state.
7. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 6, characterized in that, When the trajectory prediction module determines whether there is a deviation between the actual movement state and the preset motion state, it includes: The operating status information includes operating speed information and operating fluctuation information; The trajectory prediction module determines the preset moving speed of the current conveying equipment based on the running speed information, and determines the motion state benchmark of the current conveying equipment based on the running fluctuation information; it compares the actual moving speed in the actual moving state with the preset moving speed, and matches the moving stable state with the motion state benchmark; When the difference between the actual moving speed and the preset moving speed exceeds the preset speed deviation range, or when the moving stable state does not match the motion state benchmark, it is determined that there is a deviation between the actual moving state and the preset motion state.
8. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 7, characterized in that, When the trajectory prediction module corrects the predicted motion trajectory of the defective bottled product based on the running status information, it includes: The trajectory prediction module determines the trajectory offset state of the defective bottled product based on the running speed information, the initial motion vector, and the actual moving speed; adjusts the predicted motion trajectory of the defective bottled product according to the trajectory offset state, and updates the expected position change process of the defective bottled product in the conveying direction; and generates a corrected predicted motion trajectory based on the adjusted expected position change process.
9. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 1, characterized in that, When the rejection control module generates a rejection control command based on the target time, it includes: The rejection control module acquires the response time information of the rejection execution mechanism and determines the rejection trigger time based on the target time and the response time information. When the rejection trigger time is reached, the rejection control instruction is generated and sent to the rejection execution mechanism to control the rejection execution mechanism to perform rejection action on the defective bottled products.
10. The intelligent vision-controlled bottled product defect detection and automatic rejection system according to claim 1, characterized in that, Also includes: The rejection control module collects product status information within the rejection area and determines whether the defective bottled products have been rejected. When it is detected that the defective bottled products have not been rejected, rejection exception information is generated.
Citation Information
Patent Citations
Intelligent production line rejecting system based on image acquisition and analysis
CN119327757A
