Cigar tobacco leaf injury condition detection device and method

Abstract

The invention discloses a cigar tobacco leaf defective condition detection device and method, and relates to the technical field of cigar tobacco leaf defective detection, the cigar tobacco leaf defective condition detection device comprises a main body table top, a backlight scanning table and a rectangular fixing frame are arranged on the main body table top in parallel, and an industrial camera assembly is rotatably arranged in the area, between the backlight scanning table and the rectangular fixing frame, of the main body table top; the rectangular fixing frame is horizontally arranged in a sliding mode in the transverse direction and the longitudinal direction. The method solves the problems that in production in the traditional technology, tobacco leaves are roughly graded mainly through a manual judgment method, indexes such as the color, the oil content, the identity and the leaf structure of the tobacco leaves are easily judged through observation and touch, and subjective influences are large; and the problems of difficulty in calculation of the area proportion and the injury degree comparison of some diseases and difficulty in quick calculation of the injury area and the number of scabs are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of cigar tobacco leaf damage detection, and in particular to a cigar tobacco leaf damage detection device and method. Background Art

[0002] The planting area of ​​cigars in my country has been expanding year by year, and the cigar market has entered a stage of steady development, with a large demand for high-quality cigar tobacco raw materials. High-end handmade cigars are more dependent on the quality of tobacco raw materials. Cigars are divided into three parts: cigar wrapper, cigar wrapper, and cigar core. Among them, cigar wrapper tobacco accounts for the smallest proportion, but the unit price is the highest. In the process of rolling cigars, the cigar wrapper is spirally wound from the ignition end of the cigar to the smoking end, which will directly reflect the appearance of the cigar. Therefore, the appearance quality requirements of cigar wrapper tobacco raw materials are relatively high, and it is the most important evaluation item in the quality weighting of cigar wrapper tobacco. Among the various factors affecting the appearance quality of cigar tobacco leaves, damage (including lesions) is the indicator with the highest weight value, which has a direct impact on the value of tobacco leaves. In the process of tobacco leaf grading, different grades of tobacco leaves have different degrees of restrictions on damage. High-grade cigar wrappers often have strict requirements on the proportion of damage area. Excessive damage area will seriously affect the availability of cigar tobacco leaves, and even refuse to purchase them.

[0003] The damage mainly includes broken tobacco leaves, green spots caused by uneven maturity, spots and streaks caused by disease, and stains left by mold and rot.

[0004] The prior art discloses a patent with announcement number CN220819841U, which includes a box, a tobacco leaf image acquisition system, a lighting system, a sample detection system and an integrated electronic control system. The tobacco leaf image acquisition system and the lighting system are respectively connected to the integrated electronic control system to realize tobacco leaf image acquisition and transmission; the sample detection system is arranged inside the box for sample placement to cooperate with the tobacco leaf image detection system for shooting; the box is a rectangular hollow cavity structure with one end open, and the interior of the box is divided into an upper chamber and a lower chamber by an isolation plate, and an industrial computer database is arranged in the upper chamber for image collection and comparison; the tobacco leaf image detection system and the lighting system are arranged on the loading table of the lower chamber; an electronic control system is arranged at one end of the box for control and drive.

[0005] As the existing devices and methods are used, the shortcomings of the technology are gradually exposed, mainly in the following aspects:

[0006] First, in current production, tobacco leaves are roughly graded mainly through manual judgment. Indicators such as tobacco leaf color, oil content, identity and leaf structure are easier to judge through observation and touch, while the judgment of damage indicators is more subject to subjective influences, and the damage conditions of tobacco leaves are different, making it difficult to make an objective comparison.

[0007] Second, for some diseases, it is difficult to calculate the area percentage and the degree of damage. For example, climate spot diseases cause spots that are smaller than sesame seeds and are dotted, making it difficult to quickly calculate the damaged area and the number of spots.

[0008] In summary, the prior art obviously has inconveniences and defects in practical use, so it is necessary to improve it. Summary of the invention

[0009] In view of the defects in the prior art, the present invention solves the problem that in the traditional technology, the tobacco leaves are roughly graded mainly by manual judgment in production, wherein the indicators such as the color, oil content, identity and leaf structure of the tobacco leaves are easier to judge by observation and touch, which are subject to greater subjective influences; and the area percentage and degree of damage of some diseases are difficult to calculate, and it is difficult to quickly calculate the damage area and the number of spots.

[0010] To solve the above problems, the present invention provides the following technical solutions:

[0011] A cigar tobacco leaf damage detection device comprises a main body table, on which a backlight scanning table and a rectangular fixed frame are arranged side by side, an industrial camera assembly is rotatably arranged in the area between the backlight scanning table and the rectangular fixed frame, and the rectangular fixed frame is horizontally slidably arranged in the transverse and longitudinal directions.

[0012] As an optimized solution, a plurality of horizontally arranged spring fixing clips are arranged around the inner wall of the rectangular fixing frame, and buffer foam is arranged on the clamping surface of the spring fixing clips.

[0013] As an optimized solution, a spring rod is horizontally fixedly connected to the inner wall of the rectangular fixing frame corresponding to each of the spring fixing clips, and the telescopic end of the spring rod is connected to the spring fixing clip.

[0014] As an optimized solution, a supporting mesh or a supporting rod arranged in a cross shape is fixedly connected to the inner wall of the rectangular fixed frame.

[0015] As an optimized solution, two longitudinal movable tracks are arranged side by side on the main table top, a transverse movable track is installed between the two longitudinal movable tracks along the longitudinal sliding direction, and the rectangular fixed frame is installed on the transverse movable track along the transverse sliding direction.

[0016] As an optimized solution, the industrial camera assembly includes a camera rotating base fixedly connected to the main table top, the rotating end of the camera rotating base is fixedly connected to a bracket, the top of the bracket is fixedly connected to an industrial camera, and the lenses of the industrial camera are a wide-angle lens and a macro lens.

[0017] As an optimized solution, translucent lamp beads are fixedly provided on the main table top, and a circle of reflective rings with an offset angle is installed around the periphery of the translucent lamp beads.

[0018] As an optimized solution, a sliding linear scanner is installed between the two long sides of the backlight scanning platform.

[0019] The present invention also discloses a method for detecting the damage of cigar tobacco leaves, which comprises the following steps:

[0020] S1: Selection of cigar tobacco leaves to be tested and pre-experimental setup; removal of moldy tobacco leaves including those with pus and spores, and selection of tobacco leaf stands or fixed gauze according to the specifications and conditions of the cigar tobacco leaves;

[0021] S2: Scanning of the whole cigar leaf; Place the leaf on top of the backlit scanning table, stretch the leaf edge, and straighten the main vein;

[0022] S3: Identification of cigar tobacco leaf disease spots and injuries; after completing the tobacco leaf image scanning, the computer will identify the tobacco leaf image and start the convolutional neural network feature extraction;

[0023] S4: Transparent photography of the damaged parts of the cigar tobacco leaves; after the overall damage of the tobacco leaves is identified, the computer records the location of the larger damaged area, and the experimenter removes the tobacco leaves from the scanning platform;

[0024] S5: Analysis of the internal conditions of the damaged parts of the cigar tobacco leaves; after obtaining the light-transmitting image of the damaged parts of the tobacco leaves, the computer analyzes the internal structure of the tobacco leaves under the light transmission and identifies the uniformity of the tissue along the veins in the leaves.

[0025] As an optimized solution, in step S1, the selected tobacco leaves are counted in order, numbered leaf by leaf, the console is turned on, and the experimental method is selected, such as starting the linear scanning and partial light transmission photography of the damaged parts, setting the linear scanning step and the backlight board brightness, setting the various photography parameters of the camera, and setting the brightness of the light-transmitting lamp beads to automatic brightness. The number, processing, and number of the tobacco leaves are input into the console for the subsequent detection results of the damaged conditions.

[0026] As an optimized solution, in step S2, the backlight is turned off, the camera is rotated to be directly above the scanning platform, and a photo of the whole tobacco leaf is taken to form the edge contour lines of the tobacco leaf; the backlight is turned on, and the linear scanner scans the color of each area of ​​the whole tobacco leaf to obtain tobacco leaf images with different steps, which are combined with the tobacco leaf edge contours taken previously and spliced ​​into a single large image file by computer.

[0027] As an optimization solution, in step S3, the computer continuously adjusts the color difference and saturation of the tobacco leaf image, and performs grayscale and inversion processing to adjust the original image into multiple images with more obvious features. Subsequently, a convolutional neural network is used to recognize the original image and several images with increased recognition, and obvious color difference spots, textures, and areas that meet the characteristics of damage are identified to obtain the area, quantity and distribution of tobacco leaf damage; if the tobacco leaf is damaged over a large area, the edges of the tobacco leaf are filled with the remaining edge contours to calculate the damage area.

[0028] As an optimized solution, in step S4, a spring clip is used to fix the edge positions of the tobacco leaf in a rectangular fixing clip, and a tobacco leaf holder or a fixed mesh is selected as appropriate to support the tobacco leaf on a plane; after the fixing is completed, the camera is rotated to the side of the light-transmitting lamp bead, and the machine vision identifies the damaged positions of the tobacco leaves previously recorded by the computer; the motorized device drives the tobacco leaves in the rectangular fixing clip so that the damaged parts of the tobacco leaves are covered on the light-transmitting lamp bead, and depending on the brightness of the image obtained by the camera, the computer adjusts the brightness of the lamp bead to adapt to the parts of the tobacco leaves with different thicknesses and structural densities until the main damaged parts of the tobacco leaves are photographed.

[0029] As an optimized solution, in step S5, the parts with no obvious color difference on the surface and obvious difference under light transmission are marked, and the damaged outline of the parts in the tobacco leaf scanning image is compared to eliminate the misjudgment caused by stains on the tobacco leaf surface, and draw a distribution image of the damaged surface and internal parts of the tobacco leaf; after the tobacco leaf is inspected, the computer records the relevant data into the serial number and starts the inspection of the next tobacco leaf at the same time.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. Compared with the current subjective cigar tobacco leaf damage identification method, the present invention uses machine vision for objective comparison, can accurately measure the type, area, distribution and quantity of cigar tobacco leaf damage, can achieve horizontal comparison between different damages, and eliminate the errors caused by human observation;

[0032] 2. Compared with the current cigar tobacco leaf damage identification method, the present invention has a convolutional neural network for feature extraction, which is more accurate and faster in tobacco leaf damage identification, and can continuously enrich the model as the sample size increases, which is progressive;

[0033] 3. Compared with the current cigar leaf image recognition and identification method, the present invention has the function of light-transmitting photography, which can reflect the damage and tissue structure changes inside the tobacco leaf, reduce the misjudgment caused by stains on the surface of the tobacco leaf, and can better reflect the tobacco leaf situation than the traditional method of only identifying by surface, which has practical significance;

[0034] 4. Compared with the current cigar tobacco leaf damage identification device, this device adds programming control, which can realize automatic control of different functional components and different photography methods, so that the determination process is programmed, and the damage of tobacco leaves is calculated by computer, which simplifies the identification process and eliminates experimental errors caused by human operation;

[0035] 5. Compared with the current cigar tobacco leaf damage identification device, this device is simple to operate and has a leaf-by-leaf numbering setting, which makes the comparison between data and tobacco leaves clearer. It has the advantages of low operating difficulty and easy promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0037] Figure 1 It is a schematic diagram of the present invention in a top view direction;

[0038] Figure 2 It is a schematic diagram of the three-dimensional structure of the present invention;

[0039] Figure 3 It is a schematic diagram of the structure of the rectangular fixing frame of the present invention;

[0040] Figure 4 A schematic diagram of scanning image comparison in a comparative example of the present invention;

[0041] Figure 5 It is a schematic diagram of local lesion identification in the comparative example of the present invention;

[0042] Figure 6 It is a schematic diagram of the damage completion in the comparative example of the present invention.

[0043] In the figure: 1-main table, 2-camera rotating base, 3-industrial camera, 4-backlight scanning table, 5-line scanner, 6-light-transmitting lamp beads, 7-longitudinal axis moving track, 8-horizontal axis moving track, 9-fixed frame moving support frame, 10-rectangular fixed frame, 11-spring fixing clamp, 12-control console. DETAILED DESCRIPTION

[0044] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0045] like Figures 1 to 6As shown, the cigar tobacco leaf damage detection device includes a main body table 1, on which a backlight scanning table 4 and a rectangular fixed frame 10 are arranged side by side, and an industrial camera 3 assembly is rotatably arranged in the area between the backlight scanning table 4 and the rectangular fixed frame 10 on the main body table 1, and the rectangular fixed frame 10 is horizontally slidably arranged in the horizontal and vertical directions.

[0046] A plurality of horizontally arranged spring fixing clips 11 are arranged around the inner wall of the rectangular fixing frame 10 , and buffer foam is arranged on the clamping surface of the spring fixing clips.

[0047] A spring rod is horizontally fixedly connected to the inner wall of the rectangular fixing frame 10 corresponding to each spring fixing clip 11 , and the telescopic end of the spring rod is connected to the spring fixing clip 11 .

[0048] A supporting mesh or a supporting rod arranged in a cross shape is fixedly connected to the inner wall of the rectangular fixed frame 10 .

[0049] Two longitudinal movable tracks 7 are arranged side by side on the main table 1, and a transverse movable track 8 is installed between the two longitudinal movable tracks for longitudinal sliding. The rectangular fixed frame 10 is installed on the transverse movable track 8 for transverse sliding through the fixed frame movable support frame 9.

[0050] The industrial camera 3 assembly includes a camera rotating base 2 fixedly connected to the main table 1, a bracket fixedly connected to the rotating end of the camera rotating base 2, and an industrial camera 3 fixedly connected to the top of the bracket. The lenses of the industrial camera 3 are a wide-angle lens and a macro lens.

[0051] A light-transmitting lamp bead 6 is also fixedly provided on the main body table 1 , and a reflective ring with an angle is installed around the periphery of the light-transmitting lamp bead 6 .

[0052] A linear scanner 5 is slidably mounted between the two long sides of the backlight scanning platform 4 .

[0053] A control console 12 is also provided on the main table 1, and the two are connected through power supply and control lines. The control console 12 is a touch-screen monitor structure, connected to the device theme table with a built-in computer, installed with a programming control system, and connected to other functional components.

[0054] The console 12 is a 14-inch 1080p monitor with a ten-point touch panel, used as a built-in computer interactive window, and equipped with data transmission interfaces such as USB-A. The built-in computer is an ordinary personal PC, installed inside the theme table 1, installed with a programming control system, and connected to the Internet.

[0055] The main table top 1 is a box-shaped structure, with an internally installed control integrated circuit connected to a built-in computer, and other functional components are installed on the top of the main table top 1 of the device.

[0056] The main table top 1 is made of phenolic resin, which is hard and not easy to be oxidized or corroded.

[0057] A camera rotating base 2 is installed at the center point of the top of the device main body table 1. The structure is a truncated cone type, and a built-in linear motor can drive the camera rotating base 2 to rotate on the horizontal plane; an industrial camera 3 is installed on the top of the camera rotating base 2, and two cameras, a wide-angle camera and a macro camera, are installed. The photographic range changes with the rotation of the camera rotating base 2. A backlight scanning platform 4 is installed on one side of the camera rotating base 2. The structure is a frosted matte LED light board, the top surface is non-reflective, the lighting is uniform, and the light intensity is adjustable; scanner moving tracks are installed on both sides of the backlight scanning platform 4. The two tracks are placed in parallel, and the linear scanner 5 reciprocates along the tracks on the horizontal plane.

[0058] A light-transmitting lamp bead 6 is installed on the other side of the camera rotating base 2. A circular mirror reflection is installed on the periphery of the lamp bead to form vertically upward parallel light. The brightness of the lamp bead can be automatically adjusted. On the other side of the light-transmitting lamp bead 6, an activity area is installed. Two parallel longitudinal axis moving tracks 7 are arranged on both sides of the activity area. A transverse axis moving track 8 is installed in the two tracks. The transverse axis moving track 8 is a hollow long strip structure and can move back and forth horizontally along the two longitudinal axis moving tracks 7. Stepper motors and transmission chains are installed in the above three tracks, and fine movement in plane coordinates can be established under the programming control system. A rectangular fixed frame 10 is installed in the transverse axis moving track 8, and its base can move back and forth in the horizontal direction along the transverse axis moving track 8. A rectangular fixed frame 10 is installed on the top. The rectangular fixed frame 10 is designed as a hollow frame, and the bottom surface can be connected to multiple support rods with adjustable positions, or replaced with a hard-surface sparse woven support mesh. Spring fixing clips 11 are evenly installed in the rectangular fixed frame 10. There are multiple copies of the spring fixing clips 11 in the form of a buffer foam clip connected to a lightweight spring. The strength of the spring is adjustable to achieve universality for tobacco leaves of different specifications.

[0059] The wide-angle lens photography range can cover the entire top of the backlight scanning platform 4, and the macro lens photography range only covers the luminous area of ​​the light-transmitting lamp beads 6. The backlight scanning platform 4 is installed on one side of the camera rotating base 2. The size is 60*100cm, which can accommodate cigar tobacco leaves of any specifications. The structure is a plurality of evenly installed LED lamp tubes with a light-distributing plate. The light intensity adopts DC dimming, which does not cause stroboscopic light during shooting. The light-distributing plate is made of frosted matte texture and does not produce mirror reflection. The two long sides of the backlight scanning platform 4 are respectively installed with linear scanner 5 tracks, in duplicate, with the same motorized device built in, which can drive the linear scanner 5 to move accurately. The linear scanner 5 is a high-precision CCD scanner, and the scanning has depth of field information.

[0060] The reflective ring with an angle is used to emit uniform and parallel light. The light intensity of the lamp beads is adjustable and can reach up to 100W / cm 2, with strong light transmission ability, it can reflect the internal structure distribution of tobacco leaves. Two longitudinal axis moving tracks are installed in parallel in the vicinity of the light-transmitting lamp beads. The structure is a rectangular strip with a built-in transmission chain. A transverse axis moving track with the same structure is installed between the two moving tracks. The tracks are all controlled by stepper motors and driven by transmission chains. The displacement distance can be controlled by the number of teeth driven by the chain to achieve precise control of plane coordinates. A fixed frame moving bracket is installed inside the transverse axis moving track, which is driven to reciprocate along the track. The fixed frame moving bracket has a rotating and lifting structure, which can be adjusted by the experimenter to adjust the relative position of the rectangular fixed frame. The rectangular fixing frame is a metal square frame structure with a size of 60*100cm. It can accommodate cigar tobacco leaves of any specifications and fix them inside. There are openings at the bottom at the midpoints of the four sides, where metal thin rods can be fixed as the supporting structure of the tobacco leaves, and can also be replaced with supporting gauze. Spring fixing clips are evenly installed inside the rectangular fixing frame. The shape of the clamp is similar to that of a ticket clip, but buffer foam is installed inside to increase friction and reduce damage to the tobacco leaves. The clamp is connected to the spring rod. After clamping the tobacco leaves, the spring rods in different directions contract to flatten the tobacco leaves. The spring rod has a screw for adjusting the tension, and the force can be adjusted to ensure that the tobacco leaves are flattened without damaging them.

[0061] The present invention also discloses a method for detecting the damage of cigar tobacco leaves, which comprises the following steps:

[0062] S1: Selection of cigar tobacco leaves to be tested and pre-experimental setup; removal of moldy tobacco leaves including those with pus and spores, and selection of tobacco leaf stands or fixed gauze according to the specifications and conditions of the cigar tobacco leaves;

[0063] S2: Scanning of the whole cigar leaf; Place the leaf on top of the backlit scanning table, stretch the leaf edge, and straighten the main vein;

[0064] S3: Identification of cigar tobacco leaf disease spots and injuries; after completing the tobacco leaf image scanning, the computer will identify the tobacco leaf image and start the convolutional neural network feature extraction;

[0065] S4: Transparent photography of the damaged parts of the cigar tobacco leaves; after the overall damage of the tobacco leaves is identified, the computer records the location of the larger damaged area, and the experimenter removes the tobacco leaves from the scanning platform;

[0066] S5: Analysis of the internal conditions of the damaged parts of the cigar tobacco leaves; after obtaining the light-transmitting image of the damaged parts of the tobacco leaves, the computer analyzes the internal structure of the tobacco leaves under the light transmission and identifies the uniformity of the tissue along the veins in the leaves.

[0067] In step S1, the selected tobacco leaves are counted in order, numbered leaf by leaf, the console is turned on, and the experimental method is selected, such as starting the linear scanning and partial light transmission photography of the damaged parts, setting the linear scanning step and backlight board brightness, setting the various photography parameters of the camera, and setting the brightness of the light-transmitting lamp beads to automatic brightness. The number, processing, and number of the tobacco leaves are input into the console for the subsequent detection results of the damaged conditions.

[0068] In step S2, the backlight is turned off, and the camera is rotated to the top of the scanning platform to take a picture of the whole tobacco leaf to form the edge contour lines of the tobacco leaf; the backlight is turned on, and the linear scanner scans the color of each area of ​​the whole tobacco leaf to obtain tobacco leaf images with different steps, which are combined with the tobacco leaf edge contours taken previously and spliced ​​into a single large image file by computer.

[0069] In step S3, the computer continuously adjusts the color difference and saturation of the tobacco leaf image, and performs grayscale and inversion processing to adjust the original image into multiple images with more obvious features. Then, a convolutional neural network is used to recognize the original image and several images with increased recognition, and obvious color difference spots, lines, and areas that meet the damage characteristics are identified to obtain the area, quantity and distribution of tobacco leaf damage; if the tobacco leaf is damaged over a large area, the remaining edge contours are used to fill in the edges of the tobacco leaf in order to calculate the damage area.

[0070] In step S4, spring clips are used to fix the edges of the tobacco leaves in the rectangular fixing clips, and a tobacco leaf holder or a fixed mesh is selected as appropriate to support the tobacco leaves on a plane. After the fixing is completed, the camera is rotated to the side of the light-transmitting lamp beads, and the machine vision identifies the damaged positions of the tobacco leaves previously recorded by the computer. The motorized device drives the tobacco leaves in the rectangular fixing clips so that the damaged parts of the tobacco leaves are covered on the light-transmitting lamp beads. Depending on the brightness of the image obtained by the camera, the computer adjusts the brightness of the lamp beads to adapt to the parts of the tobacco leaves with different thicknesses and structural densities until the main damaged parts of the tobacco leaves are photographed.

[0071] In step S5, the parts with no obvious color difference on the surface and obvious difference under light transmission are marked, and compared with the damaged outline of the parts in the tobacco leaf scanning image, the misjudgment caused by stains on the tobacco leaf surface is eliminated, and the damage distribution image of the tobacco leaf surface and interior is drawn; after the tobacco leaf is inspected, the computer records the relevant data into the serial number and starts the inspection of the next tobacco leaf at the same time.

[0072] The detection process of the present invention is:

[0073] 1. Selection of cigar tobacco leaves to be tested and setting of experimental parameters

[0074] Before the experiment begins, select the cigar wrapper grade tobacco leaves purchased this year, randomly select 10 tobacco leaves of uniform size, moderate thickness, and no pus and spores, and install the supporting gauze at the bottom of the rectangular fixed frame 10. Number the selected tobacco leaves 1-10 in order, turn on the console 12, start the whole leaf scanning and the damaged part light transmission shooting, set the scanning step to 10cm per piece, the backlight brightness to automatic, the camera selects automatic exposure when shooting the outline of the tobacco leaves, and sets fixed parameters when shooting the light-transmitting tobacco leaves. Enter the number of samples for this experiment as 10 in the console 12, and the names are 1-10.

[0075] 2. Whole leaf cigar scanning

[0076] Place the tobacco leaf on the center of the top surface of the backlight scanning platform 4, with the main vein direction parallel to the long side of the backlight scanning platform 4, straighten the main vein of the tobacco leaf, stretch the leaf tip and leaf edge, and start the measurement on the console 12. The backlight of the backlight scanning platform 4 is turned off, the camera rotating base 2 rotates to the direction of the backlight scanning platform 4, and the wide-angle camera of the industrial camera 3 collects the contour image of the tobacco leaf; the backlight of the backlight scanning platform 4 is turned on, and the tobacco leaf is scanned step by step to obtain the image of each scanning segment. The built-in computer splices the high-definition image obtained by the step scanning according to the contour image of the tobacco leaf. Since the scanning parameters are fixed, the resolution of the obtained image is fixed, and the scanning range is the top of the entire backlight scanning platform 4, which remains unchanged, so the area calculation can be performed in the next step.

[0077] 3. Identification of cigar tobacco leaf spots and injuries

[0078] The built-in computer recognizes the high-definition image of the whole tobacco leaf and uses the convolutional neural network to extract the features of the tobacco leaf to identify the type, area, quantity and distribution of the damage on the tobacco leaf, and can also determine the specific area size of a certain damage.

[0079] The calculation formula for the injured area and proportion is:

[0080] Damaged area ratio = (number of pixels in the damaged part / total pixel rate of the whole tobacco leaf) * 100%

[0081] Tobacco leaf area = (total pixels of the whole tobacco leaf / total pixels on the top of the backlight scanning platform) * (100 * 60) cm 2

[0082] Total damaged area = damaged area ratio * tobacco leaf area

[0083] Single damaged area = (number of pixels in the single damaged part / total pixels on the top of the backlight scanning platform) *

[0084] (100*60)cm 2

[0085] When performing feature recognition, the computer will increase and decrease the color difference and saturation of the tobacco leaf image, and invert the original image to make the damaged features more obvious and increase the recognition accuracy. If there are no similar features in the database for tobacco leaves, they will be marked separately for subsequent experimenters to identify. At the same time, the new damaged features can be stored as training models. If the tobacco leaves are damaged over a large area, the computer will fill in the missing parts of the tobacco leaves according to the edge shape trend of the tobacco leaves and the symmetrical side shape of the main veins, and calculate the area of ​​the missing parts. According to the location of the damage and lesions, the computer can give recommendations on the availability of tobacco leaves. If the same damage appears on the edge of the leaf, it will have little effect on the availability of the wrapper tobacco leaves, but if it appears in the heart of the leaf, it may result in the tobacco leaves being used only as cigar cores.

[0086] 4. Photography of damaged cigar leaves through light

[0087] After the overall damage of the tobacco leaves is identified, the computer marks the damaged area greater than 1% or with large color difference, and the experimenter then removes the tobacco leaves from the top of the backlight scanning table 4. The experimenter fixes the tobacco leaves in the rectangular fixing frame 10, and uses the spring fixing clips 11 to flatten the tip, base, and both sides of the tobacco leaves. The tobacco leaves are supported by the gauze at this time. After the fixing is completed, the tightness of each spring fixing clip 11 is adjusted to ensure that the tobacco leaves are located in the center of the rectangular fixing frame 10. After the fixation is completed, the light-transmitting photography is turned on at the console 12, the camera rotating base 2 rotates in the direction to the light-transmitting lamp bead 6, the fixed frame moving bracket 9 moves the tobacco leaves along the longitudinal axis moving track 7 and the transverse axis moving track 8 to the top of the light-transmitting lamp bead 6, and the macro camera of the industrial camera 3 collects the image of the tobacco leaves here, and compares the whole leaf scanning image of the tobacco leaves to determine the position where the tobacco leaves are light-transmitting at this time, and establishes a plane coordinate system based on this position, the Y coordinate axis value corresponds to the movement of the longitudinal axis moving track 7, and the X axis coordinate value corresponds to the movement of the transverse axis moving track 8, and then the tobacco leaves will calculate the areas that need to be light-transmitted and take pictures one by one. After the built-in computer takes pictures at a position, it adjusts the brightness of the light-transmitting lamp bead 6 according to the brightness and feature clarity of the image until the image can clearly highlight the internal veins and organizational structure of the tobacco leaves, but the brightness is not enough to cover the tobacco leaf texture.

[0088] 5. Analysis of the internal conditions of damaged parts of cigar tobacco leaves

[0089] After the light-transmitting photography of the damaged parts of the tobacco leaves is completed, the built-in computer analyzes the obtained images, highlights the veins of the tobacco leaves in the image, and analyzes the tissue structure and uniformity of the tobacco leaves from the veins to the surrounding areas, and implements the changes in the internal structure and normal tissue around the damaged parts to determine the internal impact range of the damage. After completion, the damaged outlines shown on the tobacco leaves are matched to obtain a comparison chart of the damaged areas on the surface and inside of the tobacco leaves, and the internal area ratio is calculated separately. After the measurement of the tobacco leaf is completed, the next leaf is measured in sequence. After all the tobacco leaves are measured, the built-in computer organizes them into detailed data and images for each number.

[0090] Comparative Example

[0091] In order to evaluate the feasibility of the method and device, we used the damage detection method and device to conduct a test on the determination of tobacco leaf spots and defects in accordance with the above method:

[0092] 1. Tobacco leaf spot determination test

[0093] like Figure 4 As shown in the figure, the left side is a high-definition scanned image of tobacco leaves, and the right side is an image with the original image color difference adjusted and inverted. After the image is extracted using a convolutional neural network, the following is shown: Figure 5 As shown, the diseased spots of the identified tobacco leaves are marked with dotted lines. After comparing with the database, the computer believes that the tobacco leaves have climate spot disease. Statistics show that there are 551 climate spots on this tobacco leaf, including 34 severe spots. Using the pixel number ratio, the total area of ​​damage is 2.84%. The damage is mainly distributed in the upper left and lower right parts of the tobacco leaves. This tobacco leaf is not suitable for use as a wrapper.

[0094] 2. Tobacco leaf defect determination test

[0095] like Figure 6 As shown in the figure, the left side is a high-definition scanned image of the tobacco leaf, the middle is an image of the tobacco leaf contour extraction, and the right side is an image after the tobacco leaf edge is completed. The blue filling range in the figure is the completed area of ​​the tobacco leaf. Using the pixel number ratio, it can be obtained that the incomplete area of ​​the tobacco leaf is 7.45%, and the main distribution position is the upper left part. This piece of tobacco leaf is not suitable for use as a wrapper.

[0096] Advantages of this method and device: This method adopts an objective evaluation method, is simple to implement, has relatively low construction difficulty, accurate results, and can fill a gap in the research field of Chinese cigar raw material quality; this device is simple to operate, has a wide range of applications, and has a fast testing speed, and has great promotion significance.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A device for detecting the damage of cigar tobacco leaves, characterized in that: The invention comprises a main table (1), on which a backlight scanning table (4) and a rectangular fixed frame (10) are arranged side by side, an industrial camera (3) assembly is rotatably arranged in the area between the backlight scanning table (4) and the rectangular fixed frame (10) on the main table (1), and the rectangular fixed frame (10) is arranged to slide horizontally in the transverse direction and in the longitudinal direction. A light-transmitting lamp bead (6) is also fixedly provided on the main body table top (1), and a reflective ring with an angle is installed around the periphery of the light-transmitting lamp bead (6).

2. The cigar tobacco leaf damage detection device according to claim 1, characterized in that: A plurality of horizontally arranged spring fixing clips (11) are arranged around the inner wall of the rectangular fixing frame (10), and buffer foam is arranged on the clamping surface of the spring fixing clips.

3. The cigar tobacco leaf damage detection device according to claim 2, characterized in that: A spring rod is horizontally fixedly connected to the inner wall of the rectangular fixing frame (10) corresponding to each spring fixing clip (11), and the telescopic end of the spring rod is connected to the spring fixing clip (11); A supporting mesh or a supporting rod arranged in a cross shape is fixedly connected to the inner wall of the rectangular fixed frame (10).

4. The cigar tobacco leaf damage detection device according to claim 1, characterized in that: The industrial camera (3) assembly comprises a camera rotating base (2) fixedly connected to the main table (1); a bracket is fixedly connected to the rotating end of the camera rotating base (2); an industrial camera (3) is fixedly connected to the top of the bracket; and the lenses of the industrial camera (3) are a wide-angle lens and a macro lens.

5. A method for detecting the damage of cigar tobacco leaves, comprising the following steps: S1: Selection of cigar tobacco leaves to be tested and pre-experimental setup; removal of moldy tobacco leaves including those with pus and spores, and selection of tobacco leaf stands or fixed gauze according to the specifications and conditions of the cigar tobacco leaves; S2: Whole leaf scanning of cigar tobacco leaves: placing the tobacco leaves on top of the backlit scanning table (4), stretching the edges of the leaves and straightening the main veins; S3: Identification of cigar tobacco leaf disease spots and injuries; after completing the tobacco leaf image scanning, the computer will identify the tobacco leaf image and start the convolutional neural network feature extraction; S4: Transparent photography of the damaged parts of the cigar tobacco leaves; after the overall damage of the tobacco leaves is identified, the computer records the location of the larger damaged area, and the experimenter removes the tobacco leaves from the scanning platform; S5: Analysis of the internal conditions of the damaged parts of the cigar tobacco leaves; after obtaining the light-transmitting image of the damaged parts of the tobacco leaves, the computer analyzes the internal structure of the tobacco leaves under the light transmission and identifies the uniformity of the tissue along the veins in the leaves.

6. The method for detecting the damage of cigar tobacco leaves according to claim 5, characterized in that: In step S1, the selected tobacco leaves are counted in order, numbered leaf by leaf, the console (12) is turned on, and the experimental mode is selected, such as starting the linear scanning and the partial light transmission photography of the damaged parts, setting the linear scanning step and the backlight panel brightness, setting the various photography parameters of the camera, and setting the brightness of the light-transmitting lamp beads (6) to automatic brightness. The number, processing, and number of the tobacco leaves are input into the console (12) so as to obtain the corresponding damage detection results later.

7. The method for detecting damage to cigar tobacco leaves according to claim 5, characterized in that: In step S2, the backlight is turned off, the camera is rotated to the top of the scanning platform, and a photo of the whole tobacco leaf is taken to form the edge contour lines of the tobacco leaf; the backlight is turned on, and the linear scanner (5) scans the color of each area of ​​the whole tobacco leaf to obtain tobacco leaf images with different steps, and the images are combined with the tobacco leaf edge contours previously taken and spliced ​​into a single large image file by computer.

8. The method for detecting damage to cigar tobacco leaves according to claim 5, characterized in that: In step S3, the computer continuously adjusts the color difference and saturation of the tobacco leaf image, and performs grayscale and inversion processing to adjust the original image into multiple images with more obvious features. Subsequently, a convolutional neural network is used to recognize the original image and several images with increased recognition, and obvious color difference spots, textures, and areas that meet the characteristics of damage are identified to obtain the area, quantity and distribution of tobacco leaf damage; if the tobacco leaf is damaged over a large area, the edges of the tobacco leaf are filled with the remaining edge contours to calculate the damage area.

9. The method for detecting the damage of cigar tobacco leaves according to claim 5, characterized in that: In step S4, a spring clip is used to fix each edge of the tobacco leaf in a rectangular fixing clamp, and a tobacco leaf support or a fixing mesh is selected as appropriate to support the tobacco leaf on a plane; after the fixing is completed, the camera is rotated to the side of the light-transmitting lamp bead (6), and the damaged position of the tobacco leaf previously recorded by the computer is identified by machine vision; the motorized device drives the tobacco leaf in the rectangular fixing clamp so that each damaged part of the tobacco leaf is covered on the light-transmitting lamp bead (6), and the computer adjusts the brightness of the lamp bead according to the brightness of the image obtained by the camera to adapt to the parts of the tobacco leaf with different thicknesses and structural densities, until the main damaged parts of the tobacco leaf are photographed.

10. The method for detecting the damage of cigar tobacco leaves according to claim 5, characterized in that: In step S5, the parts with no obvious color difference on the surface and obvious difference under light transmission are marked, and the damaged outline of the parts in the tobacco leaf scanning image is compared to eliminate the misjudgment caused by stains on the tobacco leaf surface, and draw a distribution image of the damaged surface and internal parts of the tobacco leaf; after the tobacco leaf is tested, the computer records the relevant data into the serial number and starts testing the next tobacco leaf at the same time.

Images