CN111562260B - A method and device for detecting lotus root mud holes based on machine vision - Google Patents

Abstract

The invention belongs to the field of food production, and particularly relates to a lotus root mud hole detection method and device based on machine vision, which are used for detecting possible mud in lotus roots on a production line. The method comprises the steps of vertically cutting lotus roots at the position with the largest diameter and 30-35mm away from the top end to form a lotus root section and a lotus root cover, arranging linear light sources on two sides of the lotus root section for irradiation, enabling light to penetrate through meat parts and lotus root holes of the lotus roots, observing the cut surfaces, enabling the lotus root holes without sludge to be transparent and bright and uniform in color, enabling the lotus root holes with the sludge to generate obvious shadows, collecting images at the cut sections by using a CCD industrial camera, and carrying out image processing algorithms such as image preprocessing, edge detection, threshold segmentation, region growing segmentation and the like on the images to detect the existence of the sludge. The detection method and the machine vision algorithm have higher accuracy and effectiveness in sludge detection, and the lotus root mud hole detection efficiency is greatly improved.

Description

A machine vision-based method and device for detecting mud holes in lotus roots

technical field

The invention belongs to the field of food production, and in particular relates to a machine vision-based lotus root mud hole detection method and device.

Background technique

The lotus root is one of the most widely grown aquatic plants in the Asian continent. The lotus root is slightly sweet and crisp, and can be eaten raw or cooked. Inspection, cleaning, sterilization, desection and peeling, slitting and other pretreatments. The quality inspection of lotus root is very important for the subsequent processing. In addition to the inspection of appearance defects, due to the growth, harvesting, storage, transportation and other links, there is silt in the root holes of individual lotus roots, which seriously affects the subsequent processing of lotus roots, and the lotus root mud holes must be carried out one by one. Inspection, at present, is generally carried out manually, and the lotus hole is repeatedly observed. The inspection is difficult, the operation is complicated, the workload is heavy, and there are missed inspections.

SUMMARY OF THE INVENTION

In order to solve the problems of difficult and time-consuming manual detection of residual mud in the hole of lotus root, the present invention proposes a method for detecting the mud hole of lotus root based on machine vision. The lotus root is cut vertically at the end with the largest diameter and 30-35mm from the top. , to form the lotus root segment and the lotus root cover, set the line light source on both sides of the lotus root segment to irradiate, the light penetrates the fleshy part of the lotus root and the hole of the lotus root, and observe at the cut section, the hole of the lotus root without silt is transparent and bright, and the color is uniform, while the hole of the lotus root with silt The lotus root hole will produce obvious shadows, and the CCD industrial camera is used to collect images on the cutting section and perform image processing to determine whether there is mud in the lotus root hole. Specifically include the following steps:

S1. Preparation: Cut the lotus root vertically at the end with the largest diameter, 30-35mm away from the top of the lotus root, to form the lotus root segment and the lotus root cover, and place the lotus root segment horizontally on the lotus root pillow groove, facing the side where the CCD industrial camera is located. side, and align the cut surface of the lotus root section with the edge of the lotus root pillow groove;

S2. Lotus root image collection: When the lotus root conveying device transports the horizontally placed lotus root segment to the camera shooting position of the lighting box, the limit switch detects the metal induction sheet under the lotus root pillow groove, and the limit switch sends a signal to the industrial computer, and the industrial computer Send out an instruction, the lotus root conveying device stops the conveying, turn on the line light source, set the line light source at the upper and lower positions of the lotus root section, the light penetrates the fleshy part of the lotus root section and the hole of the lotus root, and the CCD industrial camera captures the image of the cut section of the lotus root;

S3. Image preprocessing: adjust the image size of the cut section of lotus root, and convert it into a grayscale image, and use Gaussian blur algorithm to remove noise and blur in the image;

S4. Edge detection: use the canny edge detection method to detect the edge of the mud in the lotus root hole.

S5. Global Adaptive Threshold Segmentation: Divide the grayscale image of the lotus root section into several sub-images, and calculate the threshold of each sub-image.

S6. Region growing segmentation: grouping pixels of the same or similar intensity into a region or spot, extracting spot features including mean and standard deviation, for identifying mud in lotus root;

S7. Determine whether there is mud in the lotus root hole.

Further, the above step S4 canny edge detection utilizes the different pixel gradients of silt and lotus root to detect the edge of silt in the lotus root hole. The magnitude and direction of the canny gradient are solved by the sobel operator, which contains two sets of 3x3 matrices, respectively Horizontally and vertically, it is convolved with the image to get the approximate value of the horizontal and vertical brightness difference respectively.

Further, the above step S5 global adaptive threshold segmentation method divides the lotus root into the background and the mud into the foreground, specifically including the following steps:

S5.1 initializes the threshold T, divides the pixel points of the lotus root section grayscale image into two categories: A class and B class;

S5.2 Calculate the mean values of the A and B pixel sets respectively.

S5.3 Calculate the inter-class variance of the two types of pixel points of A and B;

S5.4 Cycle T from 0 to 255 to calculate the inter-class variance of A and B pixels respectively. When the inter-class variance of pixel points is the largest, the corresponding T is the optimal segmentation or binarization threshold.

Further, the above S6 region growth segmentation specifically includes the following steps:

S6.1 Sequentially scan the grayscale image of the cut section of lotus root to find the first pixel that has not yet been assigned, and set the pixel to be (x ₀ , y ₀ );

S6.2 Take (x ₀ , y ₀ ) as the center, find out the four-field pixel (x, y) of (x ₀ , y ₀ ), if (x ₀ , y ₀ ) satisfies the growth rule, convert (x, y ) is merged with (x ₀ ,y ₀ ) in the same area, and (x,y) is pushed onto the stack at the same time;

S6.3 Take a pixel from the stack as a new (x ₀ , y ₀ ) and return to step S6.2.

S6.4 Return to step S6.1 when the stack is empty.

S6.5 Repeat S6.1-S6.4 until the growth ends when every point in the image has an attribution.

The present invention also proposes a lotus root mud hole detection device based on machine vision, including an industrial computer and a computer, used for image processing, signal transmission and controlling related components according to instructions, and also includes a lotus root conveying device, an image acquisition and detection device; lotus root conveying The device is used for stably and continuously transporting lotus root to the detection position in the lighting box, including lotus root pillow groove, chain with curved plate, chain track, track support plate and frame; the lotus root pillow groove is installed on the chain with curved plate, For the lotus root section placed horizontally, there is a square hole at the bottom of the lotus root pillow groove so that the light can pass through the lotus root section; the curved plate chain is installed with the lotus root pillow groove and slides in the chain track for smooth and continuous transportation of the lotus root To the detection position of the lighting box; the chain track is installed on the frame through the track support plate;

The image acquisition and detection device carries out the continuous acquisition and processing of the lotus root section image through the cooperation of the lotus root conveying device, including a lighting box, a metal induction sheet, a limit switch, a line light source, and a CCD industrial camera;

The lighting box provides a standard lighting environment for image acquisition. The lighting box is built with black linen to avoid interference from external light; the metal sensor is installed under the lotus root pillow groove, and cooperates with the limit switch to detect whether the lotus root pillow groove has been detected. position; the limit switch is installed at the detection position, and is connected with the industrial computer, and the metal induction sheet sends a signal to the industrial computer when the lotus root pillow groove reaches the limit switch, stops the conveying device, and the CCD industrial camera takes pictures; the CCD The industrial camera is placed in the lighting box, one end is connected to the industrial computer, and the detection position is facing the section of the lotus root, which is used to collect the image of the section of the lotus root and transmit the detection result to the industrial computer; the line light source is installed in the lighting box, It is connected with the industrial computer, and there is a line of light sources above and below the detection position, which is used to illuminate the lotus root with strong light.

Further, the above-mentioned image acquisition and detection device also includes a support plate, a foot bracket, a CCD industrial camera installation bracket, a support base, a square tube, and a camera mounting seat. The support plate is installed on the foot bracket to provide support for the image acquisition and detection device The CCD industrial camera mounting bracket is installed on the support base to support the line light source and the CCD industrial camera; the CCD industrial camera is installed on the CCD industrial camera mounting bracket through the camera mounting base, and the support base is installed on the The square tube is welded on the frame of the lotus root conveying device; the installation bracket is installed on the square tube, and the square tube is fixed on the frame by welding.

The present invention has following beneficial effect:

1. The present invention innovatively proposes to illuminate the fleshy part of the lotus root with strong light, collect images on the cut surface of the lotus root, and identify the silt quickly and effectively through the image processing algorithm of the silt image characteristics, which solves the problem of automatic detection of the lotus root pore silt.

2. Compared with the traditional detection method, the detection speed is fast, the efficiency is high, and it is stable and unified. The device can be applied to the industrial assembly line processing of lotus root.

Description of drawings

Fig. 1 is the structural diagram of the lotus root mud hole detection device based on machine vision of the present invention.

Fig. 2 is a partial structural schematic diagram of the lotus root mud hole detection device based on machine vision of the present invention.

Fig. 3 is a flow chart of the method of the present invention.

Fig. 4 is the image after the gray scale change of the lotus root of the present invention.

Fig. 5 is the lotus root image through the fine edge detection of the present invention.

Fig. 6 is the lotus root image through adaptive thresholding in the present invention.

Fig. 7 is an image of different lotus roots analyzed by region growing segmentation in the present invention.

Detailed ways

The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings, but the implementation and protection scope of the present invention are not limited thereto.

Fig. 1 is the structural diagram of the lotus root mud hole detection device based on machine vision of the present invention. Fig. 2 is a partial schematic diagram of Fig. 1, as shown in Fig. 1 and Fig. 2, the device includes: the lotus root conveying device includes a lotus root pillow groove 101, a chain with a curved plate 102, a chain track 103, a track supporting plate 104, and a frame 105, for smooth and continuous delivery of lotus root to the detection position in the lighting box; the image collection and detection device includes a lighting box 201, a support plate 202, a foot support 203, a metal induction sheet 204, a limit switch 205, a line light source 206, CCD industrial camera 207, camera mounting base 208, CCD industrial camera mounting bracket 209, square tube 210.

The chain with bent plate 102 is installed with lotus root pillow groove 101 and slides in the chain track 103; the track supporting plate 104 is screwed with the chain track 103 and installed on the frame 105; the lighting box 201 is installed on the support plate 202, the four corners of the support plate are equipped with foot brackets 203; the metal induction sheet 204 is welded below the lotus root pillow groove 101; the limit switch 205 is installed on the frame 105 and used in conjunction with the metal induction sheet 204 When detecting the lotus root pillow groove 101 to the detection position of the CCD industrial camera 207, stop immediately, and shoot; , used to irradiate the lotus root with strong light; the CCD industrial camera 207 is installed on the camera mount 208 to take images of the cut section of the lotus root; the camera mount 208 is mounted on the bracket 209; On the pipe 210, the square pipe 210 is fixed on the frame 105 by welding.

In this example, the lotus root is cut vertically at the end with the largest diameter, 30-35mm away from the top, to form the lotus segment and the lotus cover, and the lotus segment is placed horizontally on the lotus root pillow groove, facing the side where the CCD industrial camera is located , and align the cut surface of the lotus root with the edge of the lotus root pillow, the lotus roots pass through the lotus root mud hole detection device one by one, and the CCD industrial camera captures the image of the cut section of the lotus root and analyzes the image.

As shown in Figure 3, the present invention proposes a kind of lotus root mud hole detection method based on machine vision, specifically comprises the following steps:

S1. Preparation: Cut the lotus root vertically at the end with the largest diameter, 30-35mm away from the top of the lotus root, to form the lotus root segment and the lotus root cover, and place the lotus root segment horizontally on the lotus root pillow groove, facing the side where the CCD industrial camera is located. side, and align the cut surface of the lotus root section with the edge of the lotus root pillow groove;

S2. Lotus root image collection: When the lotus root conveying device transports the horizontally placed lotus root segment to the camera shooting position of the lighting box, the limit switch detects the metal induction sheet under the lotus root pillow groove, and the limit switch sends a signal to the industrial computer, and the industrial computer Send out an instruction, the lotus root conveying device stops conveying, turn on the line light source, set up a line light source at the upper and lower positions of the lotus root section to irradiate, the light penetrates the fleshy part of the lotus root section and the hole of the lotus root, and observe at the cutting section, the lotus root hole without silt is transparent and bright, and the color is uniform , while the lotus root hole with silt will produce obvious shadows, and the CCD industrial camera takes the image of the lotus root cut-off section;

S3. Image preprocessing: adjust the image size of the cut section of lotus root, and convert it into a grayscale image, and use Gaussian blur algorithm to remove noise and blur in the image;

图像的每一个像素的横向及纵向梯度近似值可用以下的公式结合，来计算梯度的大小：

然后可用以下公式计算梯度方向：

S4. Edge detection: Use the canny edge detection method to detect the edge of the mud in the lotus root hole. Since the pixel gradients of the mud and the lotus root are different, this algorithm can detect the edge of the mud in the lotus root hole. Canny edge detection utilizes the difference in pixel gradient between mud and lotus root to detect the mud edge in the lotus root hole. The magnitude and direction of the canny gradient are solved by the sobel operator, which contains two sets of 3x3 matrices, which are horizontal and vertical respectively. It is convolved with the image plane, and the approximate values of the horizontal and vertical brightness differences can be obtained respectively. If A represents the original image, Gx and Gy represent the images detected by horizontal and vertical edges respectively, the formula is as follows:

The horizontal and vertical gradient approximations of each pixel of the image can be combined with the following formula to calculate the magnitude of the gradient:

The gradient direction can then be calculated using the following formula:

S5. Global Adaptive Threshold Segmentation: Divide the grayscale image of the lotus root section into several sub-images, and calculate the threshold of each sub-image. The global adaptive threshold segmentation method divides the lotus root into the background and the mud into the foreground, which specifically includes the following steps:

S5.1 initializes the threshold T, divides the pixel points of the lotus root section grayscale image into two categories: A class and B class;

S5.2 Calculate the mean values of the A and B pixel sets respectively.

S5.3 Calculate the inter-class variance of the two types of pixel points of A and B;

S5.4 Cycle T from 0 to 255 to calculate the inter-class variance of A and B pixels respectively. When the inter-class variance of pixel points is the largest, the corresponding T is the optimal segmentation or binarization threshold.

S6. Region Growth Segmentation: Group pixels of the same or similar intensity into a region or spot, extract spot features including mean and standard deviation, for identifying mud in lotus root; Region Growth Segmentation specifically includes the following steps:

S6.1 Sequentially scan the grayscale image of the cut section of lotus root to find the first pixel that has not yet been assigned, and set the pixel to be (x ₀ , y ₀ );

S6.2 Take (x ₀ , y ₀ ) as the center, find out the four-field pixel (x, y) of (x ₀ , y ₀ ), if (x ₀ , y ₀ ) satisfies the growth rule, convert (x, y ) is merged with (x ₀ ,y ₀ ) in the same area, and (x,y) is pushed onto the stack at the same time;

S6.3 Take a pixel from the stack as a new (x ₀ , y ₀ ) and return to step S6.2.

S6.4 Return to step S6.1 when the stack is empty.

S6.5 Repeat S6.1-S6.4 until the growth ends when every point in the image has an attribution.

S7. Determine whether there is mud in the lotus root hole.

The parameters considered when the present invention analyzes images are: size, shape, texture and color. The color and texture of the silt and lotus root flesh have different pixel intensities, so an image segmentation process is used to distinguish them. Images are segmented into background and foreground regions. Shape, color, texture and size are mainly determined by pixel intensity gradients using canny edge detection. Resize the image to the desired size, then convert it to a grayscale image. Figure 4 shows the grayscale image obtained after running the algorithm. After converting the lotus root image into a grayscale image, a denoising algorithm is used to remove the noise and blur in the image.

Canny edge detection algorithm for detecting the boundary of lotus root hole silt. Since the pixel gradients of silt and lotus root are different, the algorithm is able to detect the edge of the silt in the hole of the lotus root. The algorithm first resizes the image and then converts it into a grayscale image. Denoise the grayscale image and determine the gradient strength of the image. The next step is to apply non-maximum suppression followed by thresholding. Figure 5 shows the results of the lotus root image obtained after edge detection analysis. A method combining adaptive global threshold segmentation and region growing segmentation is used to distinguish lotus root paste, and a new global adaptive threshold segmentation algorithm is proposed, which divides the image of lotus root into several sub-images, and calculates the threshold. Due to the different pixel intensities of the collected images, the algorithm divides the lotus root into the background and the mud into the foreground. Figure 6 is the result of adaptive threshold analysis of the lotus root image. The algorithm employed is the region growing segmentation algorithm, using color, texture, shape, size, motion, mean intensity and variance as its region-determining features, pixels adjacent to the initial seed point are inspected and checked to see if it qualifies to be part of a particular region . In order to obtain the best results as shown in Figure 7 below, good parameters should be chosen to identify the areas, i.e. the silt in the segmentation analysis. The algorithm iteratively mixes pixels or image patch regions by retaining the minimum global search cost. The process includes: using feature points as seed pixels, considering adjacent pixels, and including pixels with minimum intensity differences from feature points in the initial pixel pool middle. Computes the mean and standard deviation of pixels. Suppose there is a set of n pixels p{p1,...,pn}, on the outer boundary of the current region, then a similarity test is performed on each pixel in p to determine whether the pixel should be included in the region. A similarity measure for pixels pi ∈ p, defined as the statistical distance, where pi ∈ p has pixel intensity pi, and is the mean and standard deviation of blobs. Only pixels with the smallest statistical distance are added to the region. A predefined threshold starts here, and pixels with a statistical distance greater than the threshold will make them ineligible to be region members. When the region growth of each seed pixel in the image stops, the blob features including mean and standard deviation are extracted and used to identify the silt in lotus root pores.

The working process of a machine vision-based lotus root mud hole detection method and device will be described below.

In the normal working process, the lotus root is cut vertically at the end with the largest diameter, 30-35mm away from the top, to form the lotus root segment and the lotus root cover, and the lotus root segment is placed horizontally on the lotus root pillow groove, facing the CCD industrial camera 208 The side where the lotus root section is located, and make the cut surface of the lotus root section aligned with the edge of the lotus root pillow groove, put the lotus root section flat in the lotus root pillow groove 101, start the machine, the chain with a curved plate 102 slides in the chain track 103, and the chain with a curved plate 102 drives the lotus root pillow groove 101 installed on the bent plate to move. When the lotus root pillow groove 101 advances to the position of the CCD industrial camera 208 in the lighting box 201, the metal sensor 204 below triggers the limit switch 205, and the limit switch 205 sends a signal to the industrial computer, and the industrial computer sends an instruction to stop the chain 103 with curved plates. At this time, the cut surface of the lotus root faces the CCD industrial camera 208, and the line light sources 206 above and below the lotus root are turned on, and the light penetrates the fleshy part of the lotus root. The CCD industrial camera 208 takes images of cut sections of lotus root, and transmits the images to the computer for processing, including a series of processing such as threshold segmentation, region growth segmentation, and edge detection, and judges whether there is mud in the lotus root holes.

Claims (3)

1. A lotus root mud hole detection method based on machine vision is characterized by comprising the following steps:

s1, preparation: vertically cutting lotus roots at the position with the largest diameter and 30-35mm away from the top ends of the lotus roots to form a lotus root section and a lotus root cover, horizontally placing the lotus root section on a lotus root pillow groove, facing the side where the CCD industrial camera is located, and aligning the cut surface of the lotus root section with the edge of the lotus root pillow groove;

s2, lotus root image acquisition: when the lotus root conveying device conveys horizontally placed lotus root sections to a CCD industrial camera shooting position of the lighting box, the limit switch detects a metal induction sheet below a lotus root pillow groove, the limit switch sends a signal to an industrial personal computer, the industrial personal computer sends an instruction, the lotus root conveying device stops conveying, a linear light source is turned on, linear light source irradiation is arranged at the upper position and the lower position of the lotus root sections, light penetrates through meat parts and lotus root holes of the lotus root sections, and the CCD industrial camera shoots images of lotus root cut surfaces;

s3, image preprocessing: adjusting the image size of the lotus root section, converting the image size into a gray image, and removing noise and blur in the image by adopting a Gaussian blur algorithm;

s4, edge detection, namely detecting the edge of sludge in the lotus root hole by using a canny edge detection method;

step S4, detecting the edge of the sludge in the lotus root hole by utilizing the difference of pixel gradients of the sludge and the lotus root, solving the amplitude and the direction of the canny gradient by using a sobel operator, wherein the operator comprises two groups of 3x3 matrixes which are respectively horizontal and vertical, and performing plane convolution on the matrixes and an image to obtain horizontal and vertical brightness difference approximate values respectively;

s5, global self-adaptive threshold segmentation, namely dividing the gray level image of the lotus root cut surface into a plurality of sub-images and calculating the threshold of each sub-image;

the step S5 of global adaptive threshold segmentation is to segment lotus roots into backgrounds and segment silt into foregrounds, and specifically comprises the following steps:

s5.1, initializing a threshold T, and dividing pixel points of the gray level image of the lotus root section into two types, namely A type and B type;

s5.2, respectively calculating the mean values of the A and B pixel sets;

s5.3, calculating the inter-class variance of the A and B pixel points;

s5.4, circulating T from 0 to 255, and respectively calculating the inter-class variance of the pixel points A and B, wherein when the inter-class variance of the pixel points is maximum, the corresponding T is the optimal segmentation or binarization threshold;

s6, area growing and dividing, namely grouping pixels with the same or similar intensity into an area or a spot, extracting spot characteristics including an average value and a standard deviation, and identifying mud in the lotus roots;

the S6 region growing and dividing method specifically comprises the following steps:

s6.1, sequentially scanning the gray level image of the lotus root section, finding out a first pixel which is not belonged to, and setting the pixel as (x) ₀ ,y ₀ )；

S6.2 with (x) ₀ ,y ₀ ) As a center, find (x) ₀ ,y ₀ ) If (x), the four domain pixel (x, y) of (c) ₀ ,y ₀ ) Satisfying the growth rule, and (x, y) and (x) ₀ ,y ₀ ) Merging in the same area, and simultaneously pushing (x, y) into a stack;

s6.3 fetching a pixel from the stack as a new (x) ₀ ,y ₀ ) Returning to step S6.2;

s6.4, when the stack is empty, returning to the step S6.1;

s6.5, repeating S6.1-S6.4 until the growth is finished when each point in the image belongs to the image;

and S7, judging whether sludge exists in the lotus root holes.

2. The machine vision-based lotus rhizome mud hole detection method according to claim 1, wherein the detection method is realized based on the following detection devices: the system comprises an industrial personal computer, a lotus root conveying device and an image acquisition and detection device, wherein the industrial personal computer and the computer are used for image processing, signal transmission and control of related components according to instructions;

the lotus root conveying device is used for stably and continuously conveying lotus roots to a detection position in the lighting box and comprises a lotus root pillow groove, a chain with a bent plate, a chain track, a track supporting plate and a rack; the lotus root pillow grooves are arranged on the chain with the bent plate and used for containing horizontally placed lotus root sections, and square holes are formed in the bottoms of the lotus root pillow grooves so that light rays can penetrate through and irradiate the lotus root sections; the lotus root pillow grooves are arranged on the chain with the bent plate and slide in the chain track, and the lotus root pillow grooves are used for stably and continuously conveying lotus roots to the detection position of the lighting box; the chain track is arranged on the rack through a track supporting plate;

the image acquisition and detection device is used for continuously acquiring and processing images of the cut surface of the lotus root by matching with the lotus root conveying device and comprises an illumination box, a metal induction sheet, a limit switch, a line light source and a CCD industrial camera; the illumination box provides a standard illumination environment for image acquisition, and black linen is arranged in the illumination box to avoid interference of external light; the metal induction sheet is arranged below the lotus root pillow groove and is matched with the limit switch to detect whether the lotus root pillow groove reaches a detection position; the limit switch is arranged at the detection position and connected with the industrial personal computer, the metal induction sheet sends a signal to the industrial personal computer when reaching the limit switch along with the lotus root pillow grooves, the conveying device is stopped, and the CCD industrial camera carries out shooting; the CCD industrial camera is arranged in the lighting box, one end of the CCD industrial camera is connected with the industrial personal computer, the CCD industrial camera is opposite to the cut surface of the lotus root at the detection position and is used for collecting images of the cut surface of the lotus root and transmitting the detection result to the industrial personal computer; the line light sources are arranged in the lighting box, connected with the industrial personal computer and used for irradiating the lotus roots with strong light, and the line light sources are arranged at the upper part and the lower part of the detection position.

3. The machine vision-based lotus root mud hole detection method as claimed in claim 2, wherein said image acquisition detection device further comprises a support plate, a foot bracket, a CCD industrial camera mounting bracket, a support base, a square tube, a camera mounting seat, said support plate is mounted on the foot bracket to provide support for the image acquisition detection device; the CCD industrial camera mounting bracket is mounted on the support base and used for supporting the linear light source and the CCD industrial camera; the CCD industrial camera is mounted on the CCD industrial camera mounting bracket through the camera mounting seat, the supporting base is mounted on the square tube, and the square tube is welded on the frame of the lotus root conveying device; the mounting bracket is arranged on the square pipe, and the square pipe is fixed on the rack in a welding mode.

Images