Vehicle-mounted machine vision detection system for apple quality

Abstract

The utility model provides a vehicle-mounted machine vision detection system for apple quality, which comprises a material transmission module, a machine vision detection module, a sorting execution module and an intelligent control module, the material transmission module is used for apple transmission and self-rolling, the machine vision detection module acquires full-surface images of apples, and the sorting execution module is used for sorting the apples. The intelligent control module calls the online multi-index apple quality detection model to detect various quality indexes of to-be-detected apples, grades of the to-be-detected apples are divided according to quality detection results, a result instruction is sent to the sorting execution module, a steering engine is adjusted to drive a movable plate to incline, and the apples fall into a specified collection box; and apple sorting and grading are completed. The apple field grading device can improve the apple field grading precision and efficiency, is suitable for various orchard environments, and solves the problem of field on-line automatic sorting of apples and other sphere-like fruits and vegetables.

Description

Technical Field

[0001] The utility model relates to a vehicle-mounted machine vision detection system for apple quality. Background Art

[0002] Post-harvest apple grading is a key link in the apple harvesting stage. Apple grading can help farmers realize graded pricing and sales, effectively increasing the income of apple growers. However, currently apple harvesting and grading in my country are mostly carried out off-site, that is, after the apples are picked, they are sent to distribution centers, and large-scale apple sorting lines or manual sorting are used. The apples are then sold in grades to achieve the purpose of high-quality fruit, high prices and processing adaptation.

[0003] At present, most apple sorting and grading methods are manual sorting or large-scale fruit sorting line sorting. The primary problem with manual sorting is the rising labor costs, and manual sorting relies heavily on the experience of the operator, is subjective and inefficient. Manual apple sorting and grading mostly identify the appearance quality, including fruit diameter and shape, disease and mildew spots, large-area bumps and damage, etc. There is a lack of judgment on minor blemishes and scratches on the apple surface, apple color, and texture level. Manual sorting cannot reflect the differences between individual apples, and cannot achieve online real-time detection of apple quality. Large-scale fruit sorting lines have been initially applied in large apple-growing areas such as Shaanxi, Shanxi, and Shandong. Commercial fruit sorting lines have low single-fruit sorting costs, comprehensive functions, high detection efficiency, and good supporting facilities. However, commercial fruit sorting line detection requires multiple transfers from "orchard-sorting center-warehouse", which can easily lead to damage to apples due to improper storage and transportation, high costs of large-scale fruit sorting lines, large land area, and high operating and maintenance costs. In addition, large-scale fruit sorting lines often require the construction of supporting facilities such as cold chain transportation nearby. For apple sorting and grading in small or scattered apple producing areas, the construction cost and use cost of large-scale fruit sorting lines are relatively high, and they cannot give full play to the high-speed and low-cost advantages of non-destructive testing and intelligent grading in apple sorting and grading.

[0004] Because the field environment is more complex and the space is small, the volume of the detection equipment is required to be higher, and the detection equipment should have a certain degree of mobility to reduce the transportation of apples and reduce the damage rate of apples. However, the collection of field apple information is easily affected by factors such as light, vibration, and temperature. Therefore, when near-infrared, hyperspectral and other detection technologies that have been applied in the commercial field are used for field apple information collection, their accuracy and stability will be significantly reduced, and they are not suitable for online non-destructive testing of apple fields. At the same time, the transmission mechanisms such as fruit trays, rollers, and hydraulic conveyor troughs used in commercial fruit selection lines are large in size and have high requirements for environmental stability, and are not suitable for apple field transportation. Due to the above reasons, it is difficult to develop and produce equipment suitable for field sorting and grading of apples.

[0005] Automated sorting and grading of apples in the field can effectively reduce the cost of apple sorting, improve production efficiency, reduce the demand for human resources and reduce various errors caused by human factors, and in the long run, it can improve the competitiveness of agricultural production. Automated sorting and grading of apples in the field can ensure that apples are sorted and graded accurately and consistently, improving product quality, which is very important for meeting market demand and improving brand reputation. At the same time, sorting and grading apples in the field can realize direct transportation from "orchard to warehouse" or "orchard to market", reducing the transshipment link and thus reducing various losses of apples due to storage and transportation.

[0006] In order to solve the problem of automatic sorting and grading of apples in the field, many scholars have conducted research in this area. However, the current related devices or methods have the following deficiencies in the online automatic detection of apples in the field:

[0007] (1) Complex transmission mechanism: Existing miniaturized apple grading devices usually use complex transmission mechanisms, such as multi-stage transmission devices and roller mechanisms. These complex structural designs increase the difficulty of operation in actual use, and the maintenance and maintenance of the equipment will also be more cumbersome, increasing the cost of equipment use and the risk of failure. At the same time, complex transmission mechanisms require more sophisticated debugging, which requires more professional personnel to operate, affecting the actual use effect of the equipment.

[0008] (2) Incomplete image acquisition: Most of the current miniaturized apple grading devices use static image acquisition, which only obtains images from a few fixed angles. This image acquisition method cannot fully capture all the characteristics of apples, especially in the detection of subtle defects on the surface of apples, which is prone to missed detection and misjudgment. At the same time, statically acquired images are more easily affected by environmental factors and shooting angles, resulting in unstable image quality, affecting the accuracy and reliability of detection.

[0009] (3) Environmental factors have a great impact: The field detection environment is more complex. Ambient light, dust, temperature and humidity changes, vibration, etc. will affect the imaging quality. Such environmental factors will increase the noise in the image, making image processing and analysis more difficult. At the same time, the complexity of the field environment also increases the difficulty of equipment operation and maintenance, affecting the normal use of the equipment and detection efficiency.

[0010] (4) Low detection rate: Due to deficiencies in the design of the transmission and sorting mechanisms, the transmission and sorting speeds of the current miniaturized apple grading devices are relatively slow and cannot meet the needs of efficient grading and detection of apples in the field. In large-scale field operations, the limitation of transmission speed will affect the efficiency of the overall operation. For example, the speed inconsistency between the roller mechanism and the conveyor belt will cause the apples to become blocked or stagnant during the transmission process, affecting subsequent image acquisition and sorting operations. Slow operation will also increase the operating time and energy consumption of the equipment, reducing work efficiency.

[0011] (5) High rate of apple damage: Due to structural design issues, apples are easily hit and scratched during transportation. For example, when the roller mechanism is used to transport apples, the apples are physically damaged to varying degrees due to uneven rolling or bouncing. Especially for apples with thinner skin, collisions and scratches will seriously affect the appearance and quality of the apples, leading to a decrease in market value. The high rate of damage during the inspection process will not only affect the sorting quality, but also cause economic losses.

[0012] (6) Limited scope of application: Due to the deficiencies of (1), (2), (3), (4), and (5) above, the existing miniaturized apple grading devices have poor adaptability in practical field applications. The complex transmission mechanism and high damage rate limit the use of the equipment in different environments, especially in fields with complex terrain or harsh environmental conditions, where the operation and maintenance of the equipment are more difficult. Utility Model Content

[0013] In response to the shortcomings of the existing technology, the utility model provides a vehicle-mounted machine vision inspection system for apple quality, which solves the problems of online multi-index automatic sorting and grading of apples in the field and the high apple damage rate of other sorting mechanisms, so as to reduce the cost of apple sorting and grading in the field, increase the income of fruit farmers, and promote the diversified and high-quality development of the apple industry; at the same time, the device is also applicable to online automated field inspection of other spherical fruits and vegetables.

[0014] The utility model achieves the above technical purpose through the following technical means.

[0015] A vehicle-mounted machine vision inspection system for apple quality, comprising:

[0016] The material transmission module comprises a gap double-helix roller transmission mechanism, a driving motor and an apple transmission mechanism frame, wherein the gap double-helix roller transmission mechanism is rotatably connected to the apple transmission mechanism frame, the driving motor is installed on the apple transmission mechanism frame, and the driving motor and the gap double-helix roller transmission mechanism are connected through a synchronous belt pulley transmission system; the gap double-helix roller transmission mechanism is divided into a fruit processing area and a detection area;

[0017] The machine vision inspection module includes an image acquisition platform, which includes an industrial camera and an LED light source. The industrial camera is installed on a three-way installation platform to achieve up-down, front-back, left-right posture adjustment. The three-way installation platform is installed on an external protective cover. An LED light source is also installed on the top plate of the external protective cover. The LED light source is located on both sides above the to-be-inspected station in the transmission direction. The to-be-inspected station is located in the inspection area;

[0018] The sorting execution module comprises a sorting mechanism frame, a sorting mechanism outer groove, a sorting rubber groove, a movable plate, a shock-absorbing pipe, an execution steering gear mechanism, a collection box and a transmission motor; the sorting mechanism outer groove is fixed to the top of the sorting mechanism frame, and a gap for the movable plate to move is left between the bottom of the sorting mechanism outer groove and the sorting mechanism frame; a guide hole and a plurality of sorting holes are opened on the circumference of the sorting mechanism outer groove, the guide hole is connected with the discharge port of the material transmission module, a movable plate is installed under the sorting mechanism outer groove at the sorting hole, the bottom of the movable plate is connected with the execution steering gear mechanism, the end of the movable plate is fixed with the shock-absorbing pipe, and the end of the shock-absorbing pipe is located in the collection box; the sorting rubber groove is located in the middle of the sorting mechanism outer groove, and the sorting rubber groove is driven by the transmission motor; a plurality of grooves are processed on the outer side of the sorting rubber groove, and the number of the grooves is greater than the number of the sorting holes; an infrared sensor B is installed in each groove of the sorting rubber groove;

[0019] The intelligent control module is connected to the material transmission module, the machine vision detection module and the sorting execution module respectively to realize the management and information exchange between the modules.

[0020] In the above technical solution, the gap double-helix roller transmission mechanism includes two hollow aluminum tubes, the connecting shafts at both ends of the hollow aluminum tubes are transitionally matched with rolling bearings, and the bearing seats of the rolling bearings are fixed on the frame of the apple transmission mechanism; the outside of the hollow aluminum tubes is covered with plastic tubes, and flexible silicone guide strips are wound on the plastic tubes, and the flexible silicone guide strips are wound according to preset gap spiral lines, and the spacing of the flexible silicone guide strips gradually increases according to the apple transmission direction.

[0021] In the above technical solution, the two hollow aluminum tube shaft ends are interference fit with the driven wheel of the synchronous belt pulley transmission system and are fixed with bolts, the motor shaft of the drive motor is connected to the driving wheel of the synchronous belt pulley transmission system with a key, and each wheel system of the synchronous belt pulley transmission system is equipped with a conveyor belt.

[0022] In the above technical solution, one end of the movable plate located at the lower part of the outer groove of the sorting mechanism is set to be semicircular, and the two side edges of the movable plate at this end are of different lengths in the outer groove of the sorting mechanism, and the length of the inner side along the clockwise direction is smaller than the length of the other side.

[0023] In the above technical solution, the fruit processing area and the detection area are divided by setting up infrared sensor A. The infrared sensor A is set on the frame of the apple conveying mechanism perpendicular to the forward direction of the apples, and the distance from the end of the minimum spacing of the flexible silicone guide strip is within the range of 550-670mm.

[0024] The above technical solution also includes an equipment start button, a power supply, an alarm light, a display screen, a storage rack, a cooling fan, a carrying handle, an aviation power port, a system switch and a plug box.

[0025] Beneficial effects of the utility model:

[0026] 1. Apple is a spherical fruit with irregular shape. In traditional inspection, fruit trays are often used for static inspection. This method increases the complexity of inspection and cannot fully capture the surface image of the apple. The use of a gap double-helix roller transmission mechanism can capture the full surface image of the apple and complete automatic apple sorting and gap transmission, thereby improving the efficiency of apple grading.

[0027] 2. The utility model drives the gap double-helix roller transmission mechanism through a driving motor and a synchronous belt. The synchronous belt transmission can maintain a constant speed and precise synchronization, ensuring the stability and consistency of the transmission mechanism. Compared with chain transmission, gear transmission or other mechanical transmission methods, the synchronous belt transmission has less vibration and provides a more stable working environment. In addition, the gap double-helix roller transmission mechanism enables the apples to roll and flip continuously and smoothly during the transmission process, ensuring that the industrial camera can collect images in all directions, uniformly illuminate to avoid shadows and reflections, and improve image quality and analysis effects. In addition, the gap double-helix roller transmission mechanism can adapt to apples of different sizes, avoiding the situation where the apples are stuck or rolling stops during the transmission process, and ensuring the continuity and stability of the transmission.

[0028] 3. The utility model adopts a combination of a roulette-type sorting rubber trough, a flap, and an executive steering mechanism, which effectively ensures the stable grading and efficient detection of apples. The roulette-type sorting rubber trough can accurately guide the apples into the corresponding grading trough, while the flap and the executive steering mechanism ensure the smoothness and accuracy of the sorting process. Through this design, the equipment can significantly reduce the floor space while maintaining efficient detection, making it more suitable for various operating environments, especially orchards and farms with limited space. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a schematic diagram of the vehicle-mounted machine vision inspection system and method for apple quality described in the utility model;

[0030] Figure 2 This is a schematic diagram of the structure of the vehicle-mounted machine vision inspection system for apple quality of the utility model;

[0031] Figure 3 It is a left view of the vehicle-mounted machine vision inspection system for apple quality of the utility model;

[0032] Figure 4 It is a rear view of the vehicle-mounted machine vision inspection system for apple quality of the utility model;

[0033] Figure 5 A top view of the vehicle-mounted machine vision inspection system for apple quality described in the utility model;

[0034] Figure 6 This is a top view of the material transmission module of the utility model;

[0035] Figure 7 This is a force diagram of apples when the gap double spiral roller conveying mechanism of the utility model is in operation;

[0036] Figure 8 This is a partial view of the image acquisition platform of the utility model;

[0037] Fig. 9 This is a front view of the sorting execution module of the utility model;

[0038] Fig.10 It is a partial detail diagram of the sorting execution module of the utility model;

[0039] In the figure; 1-image acquisition platform, 2-equipment start button, 3-flexible silicone guide strip, 4-gap double spiral roller transmission mechanism, 5-rolling bearing, 6-synchronous belt pulley transmission system, 7-drive motor, 8-power supply, 9-apple transmission mechanism frame, 10-cup-shaped foot, 11-alarm light, 12-sorting mechanism outer groove, 13-sorting rubber groove, 14-lever, 15-shock-absorbing pipe, 16-collection box, 17-executor steering gear mechanism, 18-sorting mechanism Rack, 19-display screen, 20-shelf, 21-cooling fan, 22-carrying handle, 23-aviation power port, 24-system switch, 25-industrial computer, 26-plug box, 27-bolts, 28-conveyor belt, 29-plastic tube, 30-hollow aluminum tube, 31-inspection station, 32-horizontal slide rail, 33-industrial camera, 34-longitudinal slide rail, 35-camera mounting plate, 36-LED light source, 37-external protective cover, 38-transmission motor. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiment of the utility model clearer, the technical solution in the embodiment of the utility model will be clearly described below in combination with the drawings in the embodiment of the utility model, taking apples as the detection object. The described examples are part of the embodiments of the utility model, but not all of the embodiments. Based on the embodiments of the utility model, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the utility model.

[0041] Embodiment 1:

[0042] Embodiment 1 of the present utility model provides a vehicle-mounted machine vision inspection system for apple quality, such as Figure 1As shown, the machine vision inspection system includes a material transmission module, a machine vision inspection module, an intelligent control module and a sorting execution module, wherein the main body of the material transmission module is a gap double-helix roller transmission mechanism 4, and the machine vision inspection module includes an image acquisition platform 1.

[0043] The intelligent control module is connected to the material transmission module, the machine vision detection module and the sorting execution module through data cables, and the intelligent control module realizes unified management and information interaction between the modules. The intelligent control module is equipped with an industrial computer 25, which is used for the control, information recording, processing and feedback of the material transmission module, the machine vision detection module and the sorting execution module. Among them, the industrial computer 25 is mainly responsible for the adjustment of the transmission speed of the material transmission module, the signal input of the image acquisition platform 1, the processing and storage of the image information of the apple to be detected, the collection of the position information of the sorting execution module and the issuance of command information, etc., to complete the movement coordination and work unification of each module of the system. An online multi-index apple quality detection model is constructed in the industrial computer 25.

[0044] The material transmission module is as follows Figure 6As shown, a part of the material transmission module is located inside the image acquisition platform 1, and the module includes a gap double spiral roller transmission mechanism 4, a rolling bearing 5, a synchronous pulley transmission system 6, a drive motor 7, an apple transmission mechanism frame 9, a cup-shaped foot 10, a bolt 27, and a conveyor belt 28. A cup-shaped foot 10 is provided at the bottom of the apple transmission mechanism frame 9, which is used to fix the apple transmission mechanism frame 9 when the vehicle-mounted apple quality machine vision inspection system is used. The gap double spiral roller transmission mechanism 4 is composed of a flexible silicone guide strip 3, a plastic tube 29 and a hollow aluminum tube 30. The number of the flexible silicone guide strip 3, the plastic tube 29 and the hollow aluminum tube 30 are 2. The connecting shafts at both ends of the hollow aluminum tube 30 use transition fit with the rolling bearing 5 to ensure stable operation. The bearing seat of the rolling bearing 5 is fixed to the apple transmission mechanism frame 9 by bolts 27. The outside of the hollow aluminum tube 30 is covered with a plastic tube 29, and then the flexible silicone guide strip 3 is wound on the outside of the plastic tube 29 according to the pre-designed gap spiral line design. The minimum spacing of the flexible silicone guide strip 3 is a single apple diameter, which is used for later evaluation, identification and positioning, and for quality grade classification. The shaft ends of the two hollow aluminum tubes 30 are interference fit with the driven wheel of the synchronous belt pulley transmission system 6 and are fixed with bolts to ensure stable power transmission. The transmission belt 28 is set on each wheel system (driving wheel, driven wheel and tensioning wheel) of the synchronous belt pulley transmission system 6, and the motor shaft of the drive motor 7 is key-connected with the driving wheel of the synchronous belt pulley transmission system 6. The driving motor 7 is fixed on the frame 9 of the apple transmission mechanism, and the driving motor 7 is controlled by a speed regulator, which is connected to the industrial computer 25, and the running speed of the driving motor 7 is regulated according to the actual sorting rate (synchronously regulating the speed of the driving motor 7 and the transmission motor 38); in order to avoid the influence of the unstable tension of the conveyor belt 28 caused by the spacing adjustment requirements of the gap double spiral roller transmission mechanism 4, a tensioning wheel is used in the synchronous pulley transmission system 6 for regulation to ensure stable transmission requirements. The spacing of the flexible silicone guide strips 3 at the feed inlet of the gap double spiral roller transmission mechanism 4 is relatively narrow, which is used to sort the apples, and then the spacing gradually widens to speed up the transmission of apples. Through the gap double spiral rollers of the transmission mechanism, improperly placed apples will be calibrated, and the flexible silicone guide strips 3 on both sides will guide the stacked and crowded apples in the transmission process to the rear, so that before entering the image acquisition platform 1, only one apple passes through a single pitch, and the apples are sorted and transmitted at different speeds. Figure 6 The dimensions of the material transfer module are shown, which are 1270 mm long and 470 mm wide. It can be seen that the overall dimensions of the spiral roller of this size are relatively small, which is convenient for vehicle-mounted or field use. The spiral pitch is determined according to the apple detection speed requirements and the acquisition capability of the industrial camera 33. In the actual detection process, the gap double spiral roller conveying mechanism 4 can dynamically adjust the double roller spacing (by changing the spacing of the bearing seats) according to actual needs to realize the transportation of different varieties of apples or adapt to the detection of spherical fruits and vegetables of different specifications.

[0045] See also Figure 7 The double-helix roller transmission mechanism 4 is divided into a fruit processing area and a detection area. Specifically, the zoning is performed by setting infrared sensors A. Two infrared sensors A are set on the apple transmission mechanism frame 9 along a direction perpendicular to the forward direction of the apples, and the distance from one end of the minimum spacing of the flexible silicone guide strip 3 is in the range of 550-670mm. Figure 7 The stress conditions can explain that the apple can roll in the gap double spiral roller conveyor mechanism 4, and then collect multi-posture images.

[0046] The image acquisition platform 1, such as Figure 8 As shown, it includes a transverse slide 32, an industrial camera 33, a longitudinal slide 34, a camera mounting plate 35, an LED light source 36, and an external protective cover 37. The industrial camera 33 is installed on a three-way mounting platform composed of a transverse slide 32, a longitudinal slide 34, and a camera mounting plate 35, which can realize left-right, front-back, and up-down posture adjustment, so as to facilitate the camera focus and position determination after installation, and realize multi-directional adjustment; the three-way mounting platform is installed on the external protective cover 37. Two LED light sources 36 are installed on the top plate of the external protective cover 37, located on both sides above the apple inspection station 31 (located in the inspection area) in the transmission direction, to provide lighting for the machine vision inspection module, and a floodlight is used at the bottom of the light source to reduce the reflection of the apple surface caused by the direct light source to ensure the imaging quality; the incident angle of the LED light source 36 can be adjusted within the range of 0°-90° to meet the needs of different incident angles of illumination, and the LED light source 36 is installed at a position higher than the industrial camera 33 to avoid the influence of the direct light source on the image acquisition of the industrial camera. The industrial camera 33 is located above the inspection station 31, which can achieve high frame rate image acquisition, ensure that multiple apple images can be collected during the transmission process, realize the full surface image acquisition of the apple, and then achieve a more comprehensive apple surface quality assessment. The interior of the external protective cover 37 is a main frame composed of lightweight steel columns, and the exterior is protected by an aluminum alloy skin. In addition, the interior is coated with diffuse reflection material to reduce the influence of external ambient light, and make the internal light source of the machine vision inspection module more stable and soft, and reduce image acquisition shadows. The external protective cover 37 is equipped with soft light cloth and switch doors (not shown in the figure) at the feed port and discharge port in the material transmission direction, which are used to inspect and adjust the camera and prevent internal light from escaping or external light from affecting internal image acquisition, so as to obtain stable apple images.

[0047] The sorting execution module is as follows Figure 2 , 3, 4, 5, and 9, comprising a sorting mechanism frame 18, a sorting mechanism outer trough 12, a sorting rubber trough 13, a movable plate 14, a shock-absorbing pipe 15, an executive steering gear mechanism 17, a collecting box 16, and a transmission motor 38; the sorting mechanism outer trough 12 is fixed to the top of the sorting mechanism frame 18 by a column, and a gap for the movable plate 14 to move is left between the bottom of the sorting mechanism outer trough 12 and the sorting mechanism frame 18; a guide hole and a plurality of sorting holes are opened on the circumference of the sorting mechanism outer trough 12, and the guide hole is connected to the discharge port of the material transmission module through a flexible rubber tube, and the sorting A flap 14 is installed below the outer groove 12 of the sorting mechanism at the hole, and the bottom of the flap 14 is connected to the executive steering mechanism 17, and the executive steering mechanism 17 is installed on the sorting mechanism frame 18; the end of the flap 14 located at the lower part of the outer groove 12 of the sorting mechanism is set to a semicircular shape, and the lengths of the two sides of the flap 14 at this end in the outer groove 12 of the sorting mechanism are different, and the length of the side edge closer to the inside in the clockwise direction is less than the length of the other side edge. In this embodiment, the angle between the straight line where the longer side edge is located and the line connecting the center of the outer groove 12 of the sorting mechanism and the center of the semicircle of the flap 14 is 25°( Fig.10 ); the end of the movable plate 14 is fixed to the shock-absorbing pipe 15, and the end of the shock-absorbing pipe 15 is located in the collecting box 16, and the collecting box 16 is installed on the sorting mechanism frame 18; the collecting box 16 is used to receive apples of different grades; the sorting rubber groove 13 is located in the middle of the outer groove 12 of the sorting mechanism, and the center of the sorting rubber groove 13 is keyed to the output shaft of the transmission motor 38, and the transmission motor 38 is installed on the outer groove 12 of the sorting mechanism; a plurality of grooves are processed on the outer side of the sorting rubber groove 13, and the number of the grooves is greater than the number of the sorting holes; an infrared sensor B (not shown in the figure) is installed in each groove of the sorting rubber groove 13, and each infrared sensor B has an independent number for distinguishing different grooves. After receiving the apple with the electronic tag, the infrared sensor B in the groove of the sorting rubber groove 13 transmits the signal to the industrial computer 25, and then the sorting rubber groove 13 drives the apple to rotate at a constant speed around the outer groove 12 of the sorting mechanism. The industrial computer 25 estimates the time when the detected apple reaches the corresponding grade of the flap 14 based on the groove position and the rotation speed of the sorting rubber groove 13, and controls the actuator servo mechanism 17 to drive the flap 14 to flip, and the detected apple falls into the collection box 16 of the corresponding grade. The actuator servo mechanism 17 can realize the tilting of the flap 14 from 0° to a maximum of 40.7°, forming a rectangular channel with a maximum height of 140mm and a width of 100mm, which meets the passing requirements of different apples.

[0048] The transmission speed of the gap double-helix roller conveyor mechanism 4 is dynamically adjusted with the operating speed of the sorting execution module at the rear end. When too many apples pass through the detection area per unit time, the industrial computer 25 will control the transmission speed of the gap double-helix roller conveyor mechanism 4 to avoid squeezing of apples or unstable operation of the sorting execution module during the detection process. A flexible rubber tube is used at the entrance of the sorting execution module to connect with the discharge port of the gap double-helix roller conveyor mechanism 4. The flexible rubber groove is 1-2 cm lower than the discharge port. When the apple passes through the image detection area and enters the sorting execution module, it will fall into the flexible rubber tube under the action of inertia. Since the transmission speed at the end of the gap double-helix roller conveyor mechanism 4 is reduced and the inlet of the sorting execution module is designed as a gentle slope, the apple can smoothly enter the sorting execution module and proceed to the next step.

[0049] The vehicle-mounted machine vision inspection system for apple quality also includes a device start button 2, a power supply 8, an alarm light 11, a display screen 19, a storage rack 20, a cooling fan 21, a carrying handle 22, an air plug power port 23, a system switch 24 and a plug box 26. The device start button 2 is located outside the external protective cover 37 ( Figure 5 ), the aviation power port 23 and the system switch 24 are arranged outside the external protective cover 37, and the power supply 8 and the plug box 26 are arranged inside the external protective cover 37 ( Figure 4 ), the warning light 11 is located on the upper part of the external protective cover 37 ( Figure 2 ), the display screen 19 and the rack 20 are located on one side of the external protective cover 37, the cooling fan 21 and the carrying handle 22 are located on both sides of the external protective cover 37, and the cooling fan 21 and the carrying handle 22 are connected to the main frame composed of lightweight steel columns. Among them, the display screen 19 is used to display the system status, apple grading and sorting information in real time, the rack 20 is used to place the tools required for debugging the system before use, the system switch 24 controls the power on and off of the system as a whole, and the plug box 26 is the power distribution facility of the system, which is responsible for the power management of all electrical equipment.

[0050] In order to avoid the influence of mechanical vibration on the transmission of each mechanism, the gap double helical roller transmission mechanism 4, the image acquisition platform 1 and the sorting execution module are not in contact or connected with flexible materials, wherein the gap double helical roller transmission mechanism 4 is connected with connectors, bolts, etc., and the components of the image acquisition platform 1 are fixed to the columns with angle pieces and connectors. The flexible silicone guide strip 3 and the shock-absorbing pipe 15 are connected with soft rubber. The synchronous pulley, each motor, and the sorting rubber groove 13 are fixed with key connection.

[0051] The vehicle-mounted machine vision inspection system for apple quality of this embodiment is aimed at quality evaluation of the entire surface of apples. It is an online precise inspection one by one, does not rely on traditional manual evaluation, and can realize automatic grading of apples according to customized grading standards. At the same time, the system is aimed at the quality indicators of apples and other spherical fruit trees. The overall size of the system is small, and the functions of each part are modular. It can be used in orchards to carry out online multi-indicator automatic detection of apples, and can provide flexible solutions for the diversified development of the apple industry, reduce apple production costs, reduce apple damage rates and efficient sorting and grading of apples in the field.

[0052] Embodiment 2:

[0053] The implementation process of the vehicle-mounted machine vision inspection system for apple quality in the utility model is as follows:

[0054] (1) The apples to be inspected are placed in the fruit sorting area of ​​the gap double-helix roller transmission mechanism 4. After the fruit sorting is completed, before the apples to be inspected enter the inspection area of ​​the gap double-helix roller transmission mechanism 4, a state in which only a single fruit exists in the spiral line will be achieved; after entering the inspection area, the information collected by the infrared sensor A of the intelligent control module is assigned an independent electronic tag according to the order in which the apples enter the inspection area, which is used for the subsequent image classification and grading instructions belonging to the same apple.

[0055] (2) After the intelligent control module obtains the trigger information of the infrared sensor A, the industrial camera 33 of the machine vision inspection module is controlled to continuously capture images of all the transmitted apples at the inspection station 31 of the double-helix roller conveyor mechanism 4. This process will complete the continuous acquisition of multi-posture images of the apples to be inspected in the rolling transmission state in the inspection area, and then obtain multiple full-surface image information of the apples.

[0056] (3) The full-surface image information of multiple apples is sent to the online multi-index apple quality detection model. The collected images are segmented according to the time series information and the target detection model is used to preprocess the images. Then, the image classification model and regression model are used to detect the various quality indicators of the apples to be detected, and the apples to be detected are divided into grades according to the comprehensive results of various quality detections (see Table 1). The quality grades of apples are divided into special grade, first grade, second grade and other fruits. At the same time, the intelligent control module will complete the recording of the full-surface quality indicators of the apples to be detected, and put the specific quality information and grade information of the apples into the electronic tags of the corresponding apples.

[0057] Table 1 Apple quality indicators and grade classification

[0058]

[0059]

[0060] (4) The intelligent control module determines the grading information of the detected apples in each groove according to the triggering time of the infrared sensor B in each groove of the sorting rubber groove 13, and then the intelligent control module controls the transmission motor 38 in the sorting execution module, coordinates the running speed of the sorting rubber groove 13, and determines the action execution angle of the servo according to the fruit diameter and shape of the detected apples, and then sends the information to the execution servo mechanism 17 of the sorting grade position corresponding to the detected apples. When the detected apples are about to reach the grading position, the corresponding execution servo mechanism 17 moves according to the determined action execution angle, drives the flap 14 to tilt a certain angle along the short axis of the flap, and the detected apples will pass through the flap 14 into the shock absorbing pipe 15, and then enter the collection box 16; the apple sorting and grading is completed.

[0061] The vehicle-mounted machine vision inspection method for apple quality further includes: different varieties of apples have different requirements for the quality index levels, so some grading indicators (in Table 1) of the online multi-index apple quality inspection model allow users to set grading parameters according to actual apple inspection requirements, specifically including parameter indicators in terms of fruit diameter and color. The design of adjustable partial grading parameters enables the online multi-index apple quality inspection model to classify the inspected apples into different quality grades according to the actual sorting and grading requirements of the user.

[0062] The online multi-index apple quality detection model includes multiple models with different functions, and the results of multiple models are combined to comprehensively score and grade the assessed quality. Specifically, the online multi-index apple quality detection model adopts a multi-model architecture. First, the target detection model is used to locate the region of interest and classify specific defect areas. Then, the image classification model is used for defect detection and multi-index grading on the apple surface (apple texture feature detection and apple color feature detection). At the same time, the regression model is used to complete the measurement of apple size and shape feature determination (apple diameter detection, apple shape detection, and apple stem detection). In terms of specific model selection, the target detection model uses the YOLO-v7 algorithm to realize the target recognition and region of interest segmentation of the apple online image, and then transmits it to the image classification model. The image classification model uses a convolutional neural network (CNN) to complete image feature extraction, defect detection and multi-index grading. In the online multi-index apple quality detection model, a convolutional long short-term memory network (ConvLSTM) is used to process time series data (such as multi-angle images of apples rolling on a conveyor belt) to ensure that multiple images processed later belong to the same apple.

[0063] The online multi-index apple quality detection model is trained using sample images. Sample image acquisition: Use a high-resolution camera to collect sample apple images from multiple angles (e.g., 5 different angles) under a uniform industrial light source, avoiding shadows and reflections. Pay special attention to the full coverage of the sample apples by the collected sample apple images. Model input parameter section: The image size uses a standardized size (e.g., 224x224 pixels) to meet the model input requirements; the color space involved in model training is HSV. In terms of model evaluation, the evaluation indicators are accuracy, recall, precision, F1-score, etc. After the model is trained, K-fold cross validation is used to evaluate the generalization ability of the model.

[0064] The trained online multi-index apple quality detection model can be used for online multi-index detection of apples.

[0065] Embodiment 3:

[0066] To illustrate the practical application effect of this system, this example uses the vehicle-mounted machine vision inspection system for apple quality described in this utility model to complete the online multi-index automatic non-destructive inspection and grading of field apples after apple harvest in an orchard, and then compares the sorting and grading results with the actual apple grades to verify the grading accuracy of the system and method. The specific steps of this example are as follows:

[0067] Based on Example 2, a field test of the system and method was conducted using four varieties and a total of 200 sample apples. The four apple varieties are Changfu No. 2, Yantai Fuji No. 3, Yantai Fuji No. 10 and Lifu No. 1. This test first refers to the national standard GB-T10651-2008 Fresh Apple-National Standard Specification to complete the actual measurement of the apple sample grade. Then adjust the relevant parameters of the vehicle-mounted machine vision inspection system for apple quality, select some apples to be inspected and put them into the image acquisition platform 1, observe the imaging quality and grading results, and adjust the image acquisition parameters and the adjustable parameters in the online multi-index apple quality detection model according to the actual grading requirements and the actual grade of the apples to be inspected. After determining the applicable parameters under the current environment, further check the LED light source 36, the gap double spiral roller conveying mechanism 4, the sorting execution module, etc. to determine whether the equipment can operate stably and effectively, whether it is affected by external environmental factors, and after completing all equipment inspections, proceed to the next step.

[0068] Press the equipment start button 2, and make sure that before pressing the button, the safety inspection around the equipment has been completed and the equipment is in a suitable working state; start the power supply 8, and before starting the power supply, make sure that the equipment is connected to a stable power supply and that there are no abnormal conditions; start the drive motor 7 and the cooling fan 21, and before starting the drive motor 7 and the cooling fan 21, make sure that these components are in normal working condition and there are no abnormal conditions; ensure that the industrial computer 25 in the intelligent control module and the material transmission module, the machine vision detection module, and the sorting execution module have stable and good information transmission, ensure that the system self-check is in good condition, and ensure that the image acquisition platform 1 has been correctly installed and connected, and can normally receive image data from the industrial camera 33. The gap double spiral roller transmission mechanism 4, the image acquisition platform 1, and the subsequent sorting execution module are controlled by the industrial computer 25, and the system self-check is in good condition.

[0069] The apples to be inspected are placed in the fruit sorting area of ​​the gap double-helix roller transmission mechanism 4 on the apple transmission mechanism frame 9. The gap design of the flexible silicone guide strip 3 completes the fruit sorting of the apples to be inspected. Before the apples to be inspected enter the inspection area, single fruit rolling transmission will be realized. Then the gap double-helix roller transmission mechanism 4 will send the apples to be inspected into the image acquisition platform 1 one by one. The infrared sensor A in front of the inspection area completes the position arrangement according to the order in which the apples to be inspected pass. The industrial camera 33 collects multiple images of the apple surface within the next 3 seconds to obtain multiple full-surface image information of the apple.

[0070] The image acquisition platform 1 transmits the full surface image of the apple to be inspected to the online multi-index apple quality inspection model for processing. The target detection model in the model pre-processes the apple image, and then obtains the fruit diameter, fruit shape, and fruit stem through the regression model. This information will be stored in the electronic tag of the corresponding apple; the apple surface defect and texture / color feature detection is performed using the image classification model, and then the grade of the apple to be inspected is determined based on the detection results of the apple surface defect and texture / color feature and the apple diameter, fruit shape, and fruit stem; all quality indicators and grade information of the inspected apple will be stored in the electronic tag of the apple for apple sorting.

[0071] The inspected apples leave the gap double-helix roller transmission mechanism 4 in turn, and enter the gap between the outer groove 12 of the sorting mechanism and the sorting rubber groove 13 through the flexible rubber groove. The inspected apples are limited by the groove of the sorting rubber groove 13 and the infrared sensor B at that position is triggered. After obtaining the trigger information of the infrared sensor B of the sorting rubber groove 13, the intelligent control module estimates the time it takes to reach the corresponding grade of the flap 14 based on the grade information of the inspected apples in the groove, and controls the actuator servo mechanism 17 to drive the flap 14 to flip, and accurately push the apples into the corresponding shock-absorbing pipe 15 and into the corresponding collection box 16, completing the apple sorting and grading process.

[0072] After the field apple grading and sorting experiment is completed, stop the operation of each module, disconnect the main power supply of the equipment and unplug the power port 23 to ensure that the equipment is in a safe state. If the system and method have abnormal conditions during the actual operation, such as stuck apples, grading program failure, equipment failure, etc., the equipment alarm light 11 will light up to prompt the staff to deal with the abnormal situation in time. The staff can use the carrying handle 22 to move the equipment or perform maintenance operations.

[0073] The apples sorted and graded by the utility model system are compared with the apple sample grades obtained by actual measurement to obtain the grading accuracy of the utility model system in actual application. The specific results are shown in Table 2:

[0074] Table 2 Classification accuracy of the utility model system in actual application

[0075]

[0076] In this example, a vehicle-mounted machine vision inspection system for apple quality was used to conduct field tests on four different varieties of apples (Changfu No. 2, Yantai Fuji No. 3, Yantai Fuji No. 10, and Lifu No. 1), with a total of 200 sample apples. The test results show that the system performs well in multiple detection indicators and achieves high-precision online multi-index apple quality detection and grading. The vehicle-mounted machine vision inspection system and method for apple quality perform well in multiple aspects such as fruit shape, fruit diameter, bumps, surface defects, fruit stems, and overall sorting and grading, with an accuracy rate of more than 90%, and some indicators exceeding 97%. The test results show that the system can realize online multi-index automatic non-destructive inspection and grading of field apples, and can provide strong technical support for apple quality inspection and grading at the orchard site. The system can not only improve production efficiency, reduce the sorting and grading costs at the orchard site, and meet the grading needs of fruit farmers for high-quality fruits and high prices, but also has significant application effects and broad application prospects.

[0077] Although the embodiments of the present invention are described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

[0078] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0079] The series of detailed descriptions listed above are only specific descriptions of feasible embodiments of the present invention. They are not intended to limit the protection scope of the present invention. All equivalent embodiments or changes that do not deviate from the technical spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A vehicle-mounted machine vision inspection system for apple quality, characterized in that: include: The material transmission module comprises a gap double-helix roller transmission mechanism (4), a driving motor (7) and an apple transmission mechanism frame (9), wherein the gap double-helix roller transmission mechanism (4) is rotatably connected to the apple transmission mechanism frame (9), the driving motor (7) is installed on the apple transmission mechanism frame (9), and the driving motor (7) and the gap double-helix roller transmission mechanism (4) are connected via a synchronous belt wheel transmission system (6); the gap double-helix roller transmission mechanism (4) is divided into a fruit processing area and a detection area; A machine vision inspection module comprises an image acquisition platform (1), wherein the image acquisition platform (1) comprises an industrial camera (33) and an LED light source (36), wherein the industrial camera (33) is mounted on a three-way mounting platform to achieve up-down, front-back, left-right posture adjustment, wherein the three-way mounting platform is mounted on an external protective cover (37), wherein an LED light source (36) is also mounted on the top plate of the external protective cover (37), wherein the LED light source (36) is located on both sides above a station to be inspected (31) in a transmission direction, wherein the station to be inspected (31) is located in an inspection area; The sorting execution module comprises a sorting mechanism frame (18), a sorting mechanism outer groove (12), a sorting rubber groove (13), a movable plate (14), a shock-absorbing pipe (15), an execution steering gear mechanism (17), a collecting box (16) and a transmission motor (38); the sorting mechanism outer groove (12) is fixed to the top of the sorting mechanism frame (18), and a gap for the movable plate (14) to move is reserved between the bottom of the sorting mechanism outer groove (12) and the sorting mechanism frame (18); a guide hole and a plurality of sorting holes are opened on the circumference of the sorting mechanism outer groove (12), and the guide hole is connected to the discharge port of the material transmission module. A movable plate (14) is installed below the outer groove (12) of the sorting mechanism at the sorting hole, the bottom of the movable plate (14) is connected to the actuator mechanism (17), the end of the movable plate (14) is fixed to the shock-absorbing pipe (15), and the end of the shock-absorbing pipe (15) is located in the collection box (16); the sorting rubber groove (13) is located in the middle of the outer groove (12) of the sorting mechanism, and the sorting rubber groove (13) is driven by a transmission motor (38); the outer side of the sorting rubber groove (13) is processed with a plurality of grooves, and the number of the grooves is greater than the number of the sorting holes; an infrared sensor B is installed in each groove of the sorting rubber groove (13); The intelligent control module is connected to the material transmission module, the machine vision detection module and the sorting execution module respectively to realize the management and information exchange between the modules.

2. The vehicle-mounted machine vision inspection system for apple quality according to claim 1, characterized in that: The gap double spiral roller transmission mechanism (4) comprises two hollow aluminum tubes (30), the connecting shafts at both ends of the hollow aluminum tubes (30) are transitionally matched with the rolling bearings (5), and the bearing seats of the rolling bearings (5) are fixed on the frame (9) of the apple transmission mechanism; the hollow aluminum tubes (30) are externally coated with plastic tubes (29), and the plastic tubes (29) are wound with flexible silicone guide strips (3), and the flexible silicone guide strips (3) are wound according to the preset gap spiral lines, and the spacing of the flexible silicone guide strips (3) gradually increases according to the apple transmission direction.

3. The vehicle-mounted machine vision inspection system for apple quality according to claim 2, characterized in that: The shaft ends of the two hollow aluminum tubes (30) are interference-fitted with the driven wheels of the synchronous belt pulley transmission system (6) and fixed with bolts. The motor shaft of the driving motor (7) is key-connected with the driving wheel of the synchronous belt pulley transmission system (6). Each wheel system of the synchronous belt pulley transmission system (6) is provided with a conveyor belt (28).

4. The vehicle-mounted machine vision inspection system for apple quality according to claim 1, characterized in that: One end of the movable plate (14) located at the lower part of the outer groove (12) of the sorting mechanism is set in a semicircular shape, and the two side edges of the movable plate (14) at this end are located in the outer groove (12) of the sorting mechanism with different lengths, and the length of the inner side along the clockwise direction is smaller than the length of the other side.

5. The vehicle-mounted machine vision inspection system for apple quality according to claim 1, characterized in that: The fruit processing area and the detection area are divided into zones by setting infrared sensors A. Two infrared sensors A are set on the apple conveying mechanism frame (9) perpendicular to the advancing direction of the apples, and the distance from the end with the minimum spacing of the flexible silicone guide strip (3) is within the range of 550-670 mm.

6. The vehicle-mounted machine vision inspection system for apple quality according to claim 1, characterized in that: It also includes an equipment start button (2), a power supply (8), an alarm light (11), a display screen (19), a storage rack (20), a cooling fan (21), a carrying handle (22), an aviation power supply port (23), a system switch (24) and a plug box (26).

Images