A method for image correction of red date machine based on FPGA recognition

Abstract

The invention discloses a method for image correction of a red date machine based on FPGA recognition, and relates to the technical field of sorting equipment, and includes the following steps: S1: obtaining the RGB original image median filter value of the original image; S2: converting the RGB value after median filtering into HSI according to RGB2HSI; S3: averaging and saving the corresponding sampling rows of multiple images; S4: taking the average of multiple images in real time and comparing them with the saved values ​​one by one, if the averages of multiple rows have large errors, adjusting the camera exposure; S5: on the basis of S4, if it is found that there is an error in the image channel division standard line, the error may be caused by the coding signal or the chain tightness, and the error has a real-time adjustment function; S6: on the basis of S4, real-time monitoring of whether the roller is falling off. The invention is used to solve the problems that the existing equipment has insufficient automatic correction or early warning degree for the faults caused by the image brightness problem, the lack of coding signal, the tightness of the chain, and the falling off of the roller, thereby causing poor sorting effect and stuck materials.

Description

Technical Field

[0001] The present invention relates to the technical field of sorting equipment, and in particular to an image correction method for a red date machine based on FPGA recognition, so as to control the stability of the system. Background Art

[0002] In recent years, with the research and development of red date sorting technology, machine sorting has gradually replaced manual sorting. Manual sorting involves subjective judgment and low efficiency, which cannot meet market demand. Due to the fierce competition among red date sorting equipment manufacturers and the continuous improvement of customer demand, in order to stabilize market share, higher requirements are put forward for the stability of sorting equipment. At present, red date sorting equipment has several specific problems:

[0003] (1) Automatically correct the inaccurate image channel division caused by encoder signal loss and chain tightness during the operation of the red date sorting machine;

[0004] (2) Automatically adjust the brightness of the image due to light attenuation or dust based on the brightness of the template image;

[0005] (3) Warning sorting machine roller detachment caused by foreign objects. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for image correction of a red date machine based on FPGA recognition, thereby controlling the stability of the system (image channel division correction and roller shedding monitoring), which specifically includes the following steps:

[0007] Step S1: Obtaining the RGB original image median filter value of the original image;

[0008] Step S2: According to RGB2HSI, the RGB value after median filtering is converted to HSI;

[0009] Step S3: Calculate the average of the corresponding sampling rows of multiple images and save them;

[0010] Step S4: The image takes multiple images in real time to calculate the average and compares it with the saved value. If the averages of multiple rows have large errors (the system has other algorithms to remove the influence of red dates and dirt), adjust the camera exposure;

[0011] Step S5: Based on S4, if it is found that there is an error in the image channel division standard line, the error may be caused by the coding signal or the tightness of the chain, and a real-time adjustment function is provided for the error;

[0012] Step S6: Based on S4, real-time monitoring is performed to determine whether the roller is falling off.

[0013] Preferably, step S3 further comprises the following steps:

[0014] S3.1: For images of a certain size, before normal work begins, the average of multiple rows of the image is calculated based on the S component and the standard brightness determined by the measuring tool. The number of sampled rows is based on the accuracy requirements, and the sampled rows are sampled at equidistant intervals in each area of ​​the image.

[0015] S3.2: Sample row mean: Take the cumulative sum of the corresponding rows of multiple images and calculate the average. The number of sampled rows determines the number of means. After the mean is obtained, save it.

[0016] Preferably, step S4 further comprises the following steps:

[0017] S4.1: Before normal startup, eliminate the influence of red dates and dirt, read the image in real time, and compare one by one according to the interval sampling method in step S3;

[0018] S4.2: The judgment standard of the sampling row mean error is determined according to the number of error rows and the degree of error, and is set as an adjustable system parameter. When an error exists, the camera exposure is automatically adjusted.

[0019] Preferably, step S5 further comprises the following steps:

[0020] S5.1: The channel division standard line is located at the center of the roller. First, the RGB image is converted into a grayscale image, and the coordinate position of the roller is determined by the connected domain algorithm, thereby calculating the coordinates of the center point of the roller;

[0021] S5.2: According to the position of the green line marked in the color image (the channel division baseline, the number of green lines is adjusted according to the accuracy requirements), by comparing the green line position with the coordinates of the center point of the roller, the time to adjust the image acquisition is determined, and the image acquisition is based on the encoder pulse signal.

[0022] Preferably, step S6 further comprises the following steps:

[0023] S6.1: Determine whether there is a roller by the image brightness or R component, and calculate the image according to the correct channel division area;

[0024] S6.2: By calculating among the n areas, if the area value of any one or more of the middle m areas is less than the set threshold, it means that there is a fall-off, and then an alarm is sent to the upper computer, where n>m.

[0025] By using the method of the present invention, the following beneficial effects can be achieved:

[0026] (1) Automatically adjust image brightness and reduce human intervention;

[0027] (2) Automatically adjust image channel division, and further enhance machine intelligence;

[0028] (3) Automatically monitor whether the roller has fallen off to avoid a series of dangerous failures such as object jamming. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0030] Figure 1 is a flow chart of Example 1 of the present application;

[0031] Figure 2 is the image on the roller in Example 4 of the present application;

[0032] Figure 3 is the image on the roller in Example 4 of the present application;

[0033] Figure 4 This is the image on the roller in Example 5 of the present application. DETAILED DESCRIPTION

[0034] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0035] Example 1

[0036] The existing equipment does not have sufficient automatic correction or early warning for faults caused by image brightness problems, missing coding signals, chain tightness, and roller detachment, which leads to poor sorting results and object jams.

[0037] For this reason, Figure 1 As shown, the embodiment of the present application discloses a method for image correction of a red date machine based on FPGA recognition, thereby controlling the stability of the system (image offset correction and roller shedding detection and warning) according to the image. The embodiment of the present application specifically includes the following steps:

[0038] Step S1: Obtaining the RGB original image median filter value of the original image;

[0039] Step S2: According to RGB2HSI, the RGB value after median filtering is converted to HSI;

[0040] Step S3: Calculate the average of the corresponding sampling rows of multiple images and save them;

[0041] Step S4: The image takes multiple images in real time to calculate the average and compares it with the saved value. If the averages of multiple rows have large errors (the system has other algorithms to remove the influence of red dates and dirt), adjust the camera exposure;

[0042] Step S5: Based on S4, if it is found that there is an error in the image channel division standard line, the error may be caused by the coding signal or the tightness of the chain, and a real-time adjustment function is provided for the error;

[0043] Step S6: Based on step S4, monitor in real time whether the roller is falling off.

[0044] Example 2

[0045] Based on Example 1, step S3 specifically further includes the following steps:

[0046] S3.1: For an image of a certain size, before normal work begins, the average of multiple rows of the image is calculated based on the S component and the standard brightness determined by the measuring tool. The number of sampled rows is based on the accuracy requirements, and the sampled rows are sampled at equidistant intervals in each area of ​​the image;

[0047] S3.2: Sample row mean: Take the cumulative sum of the corresponding rows of multiple images and calculate the average. The number of sampled rows determines the number of means. After the mean is obtained, save it.

[0048] The technical solution in the above-mentioned embodiment 2 of the present application has at least the following technical effects or advantages:

[0049] (1) Strong anti-interference ability, eliminating interference from dust and powder;

[0050] (2) FPGA algorithm has high computational efficiency.

[0051] Example 3

[0052] Based on Example 1, step S4 specifically further includes the following steps:

[0053] S4.1: Before normal startup, eliminate the influence of red dates and dirt, read the image in real time, and compare one by one according to the interval sampling method in step S3;

[0054] S4.2: The judgment standard of the sampling row mean error is determined according to the number of error rows and the degree of error, and is set as an adjustable system parameter. When an error exists, the camera exposure is automatically adjusted.

[0055] The technical solution in the above-mentioned embodiment 3 of the present application has at least the following technical effects or advantages:

[0056] After the light source has been working for a long time, the brightness decays, causing the image to darken, which in turn affects the sorting effect. In response to this situation, the problem of increased costs such as manual maintenance is solved.

[0057] Example 4

[0058] Based on Example 1, step S5 specifically further includes the following steps:

[0059] S5.1: The channel dividing standard line is located at the center of the roller, such as Figure 2 First, the RGB image is converted into a grayscale image, and the coordinate position of the roller is determined by the connected domain algorithm, thereby calculating the coordinates of the center point of the roller;

[0060] S5.2: According to the position of the green line marked in the color image (the channel division baseline, the number of green lines is adjusted according to the accuracy requirements), by comparing the green line position with the coordinates of the center point of the roller, the time to adjust the image acquisition is determined, and the image acquisition is based on the encoder pulse signal. Figure 3 It is typical that the green line position deviates a lot from the center point of the roller, and the acquisition time is adjusted according to the deviation distance. Through such algorithm processing, the purpose of automatic adjustment and division of the camera is achieved, and the accuracy of the time of blowing out the red dates is not affected.

[0061] The technical solution in the above-mentioned embodiment 4 of the present application has at least the following technical effects or advantages:

[0062] (1) For the first time debugging, it is almost a fool-proof operation. The debugger only needs to align the camera and adjust the focal length. The red dates area is automatically divided, and the intelligence is further improved.

[0063] (2) The machine automatically corrects inaccurate divisions that occur during operation at the customer's site, reducing the frequency of after-sales staff visits.

[0064] Example 5

[0065] Based on Example 1, step S6 specifically further includes the following steps:

[0066] S6.1: Roller falls off Figure 4 As shown. The image brightness or R component is used to determine whether there is a roller, and the image is divided into regions and calculated based on the correct channel. Figure 4 It is divided into 8 areas;

[0067] S6.2: If the area value of any one or more of the 6 middle areas among the 8 areas is less than the set threshold, it means that there is a fall-off, and then an alarm is sent to the upper computer.

[0068] The technical solution in the above-mentioned embodiment 5 of the present application has at least the following technical effects or advantages:

[0069] (1) FPGA parallel processing, fast computing speed, to achieve the purpose of real-time monitoring and real-time alarm;

[0070] (2) Timely warning to avoid a series of dangerous failures such as jamming, leakage of dates and chain damage.

[0071] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for image correction of a red date machine based on FPGA recognition, characterized in that: The following steps are involved: Step S1: Obtaining the RGB original image median filter value of the original image; Step S2: According to RGB2HSI, the RGB value after median filtering is converted to HSI; Step S3: averaging the sampling rows corresponding to the multiple images and saving them, and the sampling row mean is the cumulative sum of the corresponding rows of the multiple images to calculate the mean; Step S4: taking multiple images in real time to calculate the average and compare them with the saved values ​​one by one. If the averages of multiple rows have large errors, adjust the camera exposure; Step S5: Based on S4, if it is found that there is an error in the image channel division standard line, the error is caused by the coding signal or the tightness of the chain, and a real-time adjustment function is provided for the error; The position of the channel division standard line is located at the center of the roller. First, the RGB image is converted into a grayscale image, and the coordinate position of the roller is determined by the connected domain algorithm, so as to calculate the coordinates of the center point of the roller. According to the position of the green line marked in the color image, by comparing the position of the green line with the coordinates of the center point of the roller, the time to adjust the image acquisition is determined, and the image acquisition is based on the encoder pulse signal. Step S6: Based on S4, real-time monitoring is performed to determine whether the roller is falling off.

2. The image correction method for a red date machine based on FPGA recognition according to claim 1 is characterized in that: The step S3 further comprises the following steps: S3.1: For images of a certain size, before normal work begins, the average of multiple rows of the image is calculated based on the S component and the standard brightness determined by the measuring tool. The number of sampled rows is based on the accuracy requirements, and the sampled rows are sampled at equidistant intervals in each area of ​​the image. S3.2: The number of sampled rows determines the number of means, which are then saved after being obtained.

3. The image correction method for red date machine based on FPGA recognition according to claim 1 is characterized in that: The step S4 further comprises the following steps: S4.1: Before normal startup, eliminate the influence of red dates and dirt, read the image in real time, and compare them one by one according to the interval sampling method in step S3; S4.2: The judgment standard of the sampling row mean error is determined based on the number of error rows and the degree of error. It is set as an adjustable system parameter and the camera exposure is automatically adjusted when an error exists.

4. The image correction method for red date machine based on FPGA recognition according to claim 1 is characterized in that: The step S6 further comprises the following steps: S6.1: Determine whether there is a roller by the image brightness or R component, and calculate the image according to the correct channel division area; S6.2: By calculating among n areas, if the area value of any one or more of the middle m areas is less than the set threshold, it means that there is a fall-off, and then an alarm is sent to the upper computer, where n>m.

Images