Visual control system and method for plastic bottle sorting based on visual inspection

Abstract

The invention relates to the technical field of visual inspection, and particularly discloses a visual control system and method for plastic bottle sorting based on visual inspection, and the method comprises the following steps: S1, starting a camera when a plastic bottle is detected, and calculating the imaging time; s2, obtaining an exposure value of the imaging picture, presetting an exposure interval, marking abnormal pixel points, calculating the proportion of the abnormal pixel points, and carrying out preliminary screening on the plastic bottles based on the proportion of the abnormal pixel points; s3, marking to-be-sorted bottles, arranging an NIR sensor, emitting near-infrared light when the to-be-sorted bottles pass, acquiring a spectrum reflected by the to-be-sorted bottles, and determining wavelengths of absorption peaks corresponding to the to-be-sorted bottles based on the spectrum; and S4, the materials of the bottles to be sorted are determined based on the wavelengths, the total time and distance coefficient are calculated, and the plastic bottles are blown out through the corresponding air valves to complete sorting. The sorting efficiency is improved, the labor cost is reduced, and meanwhile, the false drop rate can be further reduced through automatic work of a machine.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual detection, in particular to a visual control system and method for plastic bottle sorting based on visual detection. BACKGROUND

[0002] In the specific field of plastic product recycling, especially in the sorting link of plastic bottles, there are a series of significant industry pain points. For a long time, the traditional way of plastic bottle sorting has faced many severe challenges, which has seriously restricted the efficient development and sustainable operation of the entire industry.

[0003] In terms of sorting efficiency, the traditional mode is extremely low. Usually relying on a large number of manual visual identification and manual sorting, workers need to distinguish the characteristics of material, color, shape, etc. in various plastic bottles one by one, and then classify and place them. This process is extremely tedious and slow, because the speed of each person's operation is limited, and even if a large number of manpower is invested, it is difficult to keep up with the pace in the face of a large number of plastic bottles. The labor cost is high. Moreover, due to the high strength and poor environment of plastic bottle sorting work, the mobility of personnel is high, and enterprises also need to continuously invest resources in the recruitment and training of new employees, further increasing the labor cost burden. And the workers' working environment is poor, long-term inhalation of harmful gases may cause respiratory diseases and other health problems. And long time high intensity work state will lead to high misjudgment rate. In continuous work, visual fatigue is inevitable. Therefore, it is easy to make mistakes and put plastic bottles that should be classified into other categories. SUMMARY

[0004] The purpose of the present application is to provide a visual control system and method for plastic bottle sorting based on visual detection, which solves the above technical problems.

[0005] The purpose of the present application can be achieved by the following technical solutions: The visual control method for plastic bottle sorting based on visual detection comprises the following steps: S1: start the camera when a plastic bottle is detected on the conveyor belt, obtain the image height H of the plastic bottle, the image height H of the plastic bottle represents the number of single row pixels collected and spliced by the camera, calculate the imaging time T = H / FL, wherein FL represents the row frequency of the camera; S2: sequentially perform the following operations on the imaging picture: obtain the exposure value E of each pixel point in the imaging picture, pre-set the exposure interval [E min , E max ], mark the pixel points with exposure value E outside the exposure interval as abnormal pixel points, wherein E min , E max represent the preset minimum and maximum exposure values respectively; Calculate the proportion of abnormal pixels B = N1 / N all N1 represents the number of abnormal pixels, N all The number of pixels occupied by plastic bottles in the image is represented by the proportion B of abnormal pixels, which is used to initially screen plastic bottles. S3: Record the plastic bottles that have passed the initial screening as bottles to be sorted. Set up an NIR sensor at a preset distance L meters from the camera. The NIR sensor includes a transmitter and a receiver. When the bottle to be sorted passes directly below the transmitter, it emits near-infrared light and obtains the spectrum reflected by the bottle. Based on the spectrum, determine the wavelength C of the corresponding absorption peak of the bottle to be sorted. S4: Determine the material of the bottle to be sorted based on wavelength C. Once the material is confirmed, obtain the time t1 when near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined, and calculate the total time T. all =T+t2-t1, calculate the distance coefficient S=L+T all ×v, where v represents the conveyor belt speed, and the air valve corresponding to the distance coefficient S is activated to blow out plastic bottles made of PET and complete the sorting.

[0006] As a further aspect of the present invention: in step S1, a minimum exposure time T is preset. min If the line frequency FL ≤ 1 / T min The line frequency FL is increased by a preset gradient coefficient λ×f until the line frequency FL > 1 / T. min , where λ represents the preset gradient value, λ=1,2,...,f represents the preset minimum row frequency.

[0007] As a further aspect of the present invention: in step S2, the method for preliminary screening of plastic bottles based on the proportion B of abnormal pixels includes: Plastic bottles with an abnormal pixel ratio B ≥ 10% are blown off the conveyor belt through an air valve, and the average exposure value E is calculated. ave The mean E ave >E max Plastic bottles that are overexposed are classified as overexposed plastic bottles, with a mean E. ave <E max The plastic bottles that are overexposed are marked as low-exposure plastic bottles. After one round of sorting, the light intensity is reduced by 10% from the original intensity, and the overexposed plastic bottles are put back into the conveyor belt for sorting. The above operation is repeated until there are no more overexposed plastic bottles. Increase the light intensity by 10% on top of the original intensity, and put the low-exposure plastic bottles back into the conveyor belt for sorting. Repeat the above operation until there are no more low-exposure plastic bottles.

[0008] As a further aspect of the present invention: in step S3, the method for determining the wavelength C of the absorption peak corresponding to the bottle to be sorted based on the spectrum includes: A preset judgment threshold is set. If the absolute value of the absorption peak exceeds the judgment threshold, the corresponding wavelength range is recorded as wavelength C.

[0009] As a further aspect of the present invention: in step S4, the method for determining the material of the bottle to be sorted based on wavelength C includes: Set a preset confidence level z, obtain the wavelength Z corresponding to the absorption peak of PET material, and set the wavelength floating range [ZZ]. sta / z,Z+Z sta If the overlap between wavelength C and the wavelength floating range exceeds 90%, then the material of the bottle to be sorted is determined to be PET.

[0010] As a further aspect of the present invention: in step S3, the maximum imaging time T is obtained. max Ensure distance L > T max ×v.

[0011] As a further aspect of the present invention: if the overlap between wavelength C and other wavelength floating ranges (excluding PET materials) exceeds 50%, then the subsequent operation is stopped, and the corresponding plastic bottle is re-sorted.

[0012] A vision-based plastic bottle sorting and selection vision control system includes: Imaging module: When a plastic bottle is detected on the conveyor belt, the camera is activated to acquire the image height H of the plastic bottle. The image height H of the plastic bottle represents the number of pixels in a single row that the camera captures and stitches together. The imaging time T is calculated as T = H / FL, where FL represents the line frequency of the camera. Preliminary screening module: Perform the following operations on the images in sequence: Obtain the exposure value E of each pixel in the image, and preset the exposure range [E]. min E max Pixels whose exposure value E is outside the exposure range are marked as abnormal pixels, where E min E max These represent the preset minimum and maximum exposure values, respectively; Calculate the proportion of abnormal pixels B = N1 / N all N1 represents the number of abnormal pixels, N all The number of pixels occupied by plastic bottles in the image is represented by B, which is used to initially screen plastic bottles based on the proportion of abnormal pixels. The material identification module: the plastic bottles screened by the preliminary screening are recorded as bottles to be sorted, an NIR sensor is arranged at a preset distance L meters away from the camera, the NIR sensor comprises an emitter and a receiver, near-infrared light is emitted when the bottle to be sorted passes directly below the emitter, the spectrum reflected by the bottle to be sorted is acquired, and the wavelength C of the corresponding absorption peak of the bottle to be sorted is determined based on the spectrum; The sorting module: the material of the bottle to be sorted is determined based on the wavelength C, when the material confirmation is completed, the time t1 when the near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined are acquired, and the total time T is calculated all =T+t2-t1, the distance coefficient S=L+T is calculated all ×v, wherein v represents the speed of the conveying belt, the distance coefficient S corresponds to the starting of the air valve to blow out the plastic bottles with the material PET and complete the sorting.

[0013] The beneficial effects of the present application are: first, the plastic bottles to be sorted are photographed by the camera, since the camera needs time to take pictures, and the conveying belt will not stop during the imaging process, therefore, the purpose of calculating the imaging time here is to adjust the air valve, so that the air valve blows up the corresponding plastic bottles for sorting.

[0014] Then the imaging picture is processed, and the exposure value of each pixel point in the imaging picture is used for judgment, it should be noted that, in normal cases, in order to make the plastic bottle sorting more accurate, light supplement measures will be taken, but various reasons may cause abnormal pixel points in the imaging picture, which will affect the subsequent identification, therefore, the exposure value of the image is used for judgment, if too many abnormal pixel points appear in an imaging picture, the subsequent operation will not be able to identify, thereby affecting the accuracy.

[0015] The plastic bottles screened by the preliminary screening are recorded as bottles to be sorted, and the bottles to be sorted are screened again by the NIR sensor, the principle here is that the absorption rate of different wavelengths of light by different bonds in chemical equations is used for distinguishing, the absorption rate of the existing materials can be used for judgment, although there may be some differences, but the different materials can still be distinguished relatively accurately, it should be noted that the present application is mainly used for recycling plastic bottles with PET material, but is not limited to this material, since different materials have different absorption intensities of light spectrum, this method can still be used to distinguish plastic bottles with different materials, and different materials of plastic bottles can be blown out in batches by controlling the starting time of the air valve.

[0016] Finally, the material of the bottle to be sorted is determined by wavelength, the time consumed in the process of imaging to determine the material of the plastic bottle is counted, and then the distance coefficient is calculated according to the speed of the conveying belt, the greater the distance coefficient, the later the air valve blowing time, the smaller the value, the earlier the blowing time, so as to complete a whole visual control process for sorting plastic bottles based on visual detection. As described above, the application not only improves the sorting efficiency and reduces the labor cost, but also utilizes the machine automation work to further reduce the misjudgment rate. BRIEF DESCRIPTION OF DRAWINGS

[0017] The application will be further described below in combination with the drawings.

[0018] Figure 1 It is a flowchart of the visual control method for sorting plastic bottles based on visual detection. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the application will be clearly and completely described below in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0020] Please refer to Figure 1 The application is a visual control method for sorting plastic bottles based on visual detection, which comprises the following steps: S1: when a plastic bottle is detected on the conveying belt, start the camera to obtain the image height H of the plastic bottle, the image height H of the plastic bottle represents the number of single-row pixels collected and spliced by the camera, and the imaging time T is calculated as T=H / FL, wherein FL represents the row frequency of the camera; S2: sequentially perform the following operations on the imaging picture: obtain the exposure value E of each pixel point in the imaging picture, pre-set the exposure interval [E min , E max ], mark the pixel points with exposure values E outside the exposure interval as abnormal pixel points, wherein E min , E max respectively represent the preset minimum exposure value and maximum exposure value; calculate the proportion B of abnormal pixel points as B=N1 / N all , N1 represents the number of abnormal pixel points, and N all represents the number of pixel points occupied by the plastic bottle in the imaging picture, and the plastic bottle is preliminarily screened based on the proportion B of abnormal pixel points; S3: The plastic bottles that pass the preliminary screening are marked as bottles to be sorted, and an NIR sensor is set at a preset distance L meters away from the camera, which includes a transmitter and a receiver. When the bottle to be sorted passes directly below the transmitter, near-infrared light is emitted, and the spectrum of the reflected light is obtained. Based on the spectrum, the wavelength C of the corresponding absorption peak of the bottle to be sorted is determined. S4: Based on the wavelength C, the material of the bottle to be sorted is determined. When the material confirmation is completed, the time t1 when the near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined are obtained, and the total time T is calculated all =T+t2-t1, and the distance coefficient S=L+T is calculated. all ×v, where v represents the speed of the conveyor belt. The distance coefficient S corresponds to the start of the air valve to blow out the plastic bottles with PET material and complete the sorting.

[0021] It should be noted that the camera is used to capture images of the plastic bottles to be sorted placed on the conveyor belt. All subsequent sorting operations depend on the accuracy and timeliness of the acquired images. It should be noted that the camera cannot complete a complete photographing process in an instant, but needs a certain amount of time to process a series of imaging operations. At the same time, during the entire imaging process, the conveyor belt does not stop running because the camera is working. The conveyor belt always maintains a stable speed to continuously convey the plastic bottles, ensuring the efficiency and continuity of the entire production process.

[0022] In this case, the imaging time is particularly critical. The core purpose is to accurately adjust the working state of the air valve according to the calculated time parameter. When the camera captures a clear image of a specific plastic bottle, the control system will send a control signal to the corresponding air valve in a timely manner according to the pre-set program and algorithm combined with the calculated imaging time. In this way, the air valve can blow the corresponding plastic bottle at the right moment, making it leave the original conveying track and be guided to different collection areas according to the predetermined path, thereby achieving efficient and accurate sorting of different types or colors of plastic bottles. This air valve adjustment mechanism based on imaging time not only improves the efficiency of sorting, but also greatly improves the accuracy of sorting, providing strong technical support for the entire plastic bottle recycling process.

[0023] After completing the photographing and collection of the plastic bottles, the next step is the image processing stage. The core task is to analyze and process the acquired imaging pictures, and the key basis is the exposure value of each pixel point in the picture.

[0024] In actual production environment, in order to ensure that the plastic bottle sorting can achieve higher accuracy, usually will take special light supplement measures. Because in natural light conditions, it is often difficult to ensure that each plastic bottle can get uniform, adequate light irradiation, and then affect the quality of the image. Through the reasonable setting of light supplement equipment, can provide stable and intensity suitable light source for plastic bottle from different angles, so that the image taken has better contrast and clarity, highlights the feature details of the plastic bottle, however, although the light supplement measures are taken, but in the specific shooting process, still will be affected by various complex factors, such as the problem caused by factors such as reflection, will be directly reflected on the imaging picture, as the exposure value of some area pixel point is abnormal.

[0025] In this case, it is necessary to judge according to the exposure value of the image. If there are too many abnormal pixels in an imaging picture, that is, the exposure value of those pixels is too high or too low, deviating from the normal range, which will bring trouble to the subsequent operation. The subsequent recognition algorithm is usually designed and run based on the normal exposure value distribution, and too many abnormal pixels will disturb the original data pattern, so that the algorithm can not accurately extract the effective feature information of the plastic bottle. Once this happens, the subsequent operation will not be able to accurately identify the type, category or other key attributes of the plastic bottle, and then seriously affect the accuracy and reliability of the whole sorting process.

[0026] After a series of preliminary screening processes, those plastic bottles that meet the basic conditions will be marked as bottles to be sorted and enter the next stage of fine identification. At this time, NIR sensor is used to carry out secondary screening on the bottles to be sorted. The core principle of this process comes from the unique absorption characteristics of different chemical bonds to specific wavelength light in the field of chemistry. In the molecular structure of matter, the chemical bonds formed between various atoms are like a "filter" with selectivity, which will show different absorption rates to light of different wavelengths. Each kind of plastic material has its own unique chemical bond combination in its molecular structure, which corresponds to a unique spectral absorption curve. When we use NIR sensor to emit light beams covering multiple wavelength ranges to irradiate the plastic bottle, the chemical bonds in the plastic bottle will selectively absorb some wavelengths of light according to their own characteristics, while other wavelengths of light will be transmitted or reflected. By precisely detecting the intensity change of the transmitted or reflected light, the characteristic spectral information of the plastic bottle material can be obtained.

[0027] When the NIR sensor collects the spectral response of the bottle to be sorted, it is compared and matched with the standard spectrum in the database. Although in real working conditions, due to process fluctuations in the production process, differences in impurity content of raw materials, and environmental erosion during use, there may be some deviation between the actual measured value and the theoretical value, but with the characteristics of NIR technology, quite accurate material differentiation can still be achieved.

[0028] It is worth emphasizing that the original design of the application is focused on recycling plastic bottles made of PET material. However, this does not mean that the scope of application of this technology is limited to PET material. Because different materials of plastic have significant differences in spectral absorption intensity, this sorting method based on NIR sensor has wide versatility. Whether it is PP, PE, or PVC and other common plastic materials, they can be effectively identified by this method.

[0029] In order to achieve efficient sorting of plastic bottles of multiple materials, the system skillfully uses a gas valve control strategy. According to the material identification results fed back by the NIR sensor in real time, the control system will accurately calculate and set the time interval for the gas valve to start. When a plastic bottle of a certain material reaches the predetermined position, the corresponding gas valve will be opened in time to generate a high-speed airflow to accurately blow out the batch of plastic bottles, so that they enter the special collection channel.

[0030] In order to realize the fine management and continuous optimization of the entire sorting process, the system will count the time consumed by the process. At the same time, combined with the speed parameter of the stable operation of the conveyor belt, the distance coefficient is calculated, which is closely related to the starting time of the gas valve: when the distance coefficient presents a larger value, it means that the system determines that the plastic bottle has not yet completely reached the best blowing position. Therefore, the blowing time of the gas valve will be correspondingly delayed to ensure that the airflow is applied at the most appropriate moment, so that the plastic bottle can be accurately blown off the conveyor belt; on the contrary, if the distance coefficient is small, it means that the plastic bottle is close to the ideal blowing point, and the system will trigger the gas valve action in advance to ensure the timeliness and accuracy of the operation.

[0031] In another preferred embodiment of the application, the minimum exposure time T min is preset, and if the line frequency FL is less than or equal to 1 / T min , the line frequency FL is increased by a preset gradient coefficient λxf until the line frequency FL is greater than 1 / T min , where λ represents a preset gradient value, λ = 1, 2,..., and f represents a preset minimum line frequency.

[0032] It is worth noting that if the bottle moves too fast and the camera exposure time is too long, the light will move a distance during the exposure time, causing the image to be blurred. Therefore, it is necessary to increase the line frequency. The higher the line frequency, the shorter the maximum exposure time allowed, because more lines are scanned per second and each line must be shorter. Short exposure eliminates trailing.

[0033] In another preferred embodiment of the present application, the method for preliminarily screening plastic bottles based on the proportion B of abnormal pixel points comprises: Plastic bottles corresponding to a proportion B of abnormal pixel points greater than or equal to 10% are blown out of the conveyor belt by the air valve, and the average value E of the exposure value is calculated ave , the average value E ave of the exposure value is calculated max , and the plastic bottles with an average value E ave greater than E max are recorded as overexposed plastic bottles, and the plastic bottles with an average value E ave less than E max are recorded as underexposed plastic bottles. After one round of sorting is completed, the original light intensity is reduced by 10%, the overexposed plastic bottles are again put into the conveyor belt for sorting, and the above operations are repeated until there are no overexposed plastic bottles. The original light intensity is increased by 10%, the underexposed plastic bottles are again put into the conveyor belt for sorting, and the above operations are repeated until there are no underexposed plastic bottles.

[0034] It can be understood that although the light supplementing measures are taken, in the specific shooting process, various complex factors such as reflection and other factors will directly affect the imaging picture, resulting in abnormal exposure values of pixel points in some areas.

[0035] In another preferred embodiment of the present application, the method for determining the wavelength C of the absorption peak corresponding to the bottle to be sorted based on the spectrum comprises: A judgment threshold is preset, and if the absolute value of the absorption peak exceeds the judgment threshold, the wavelength range corresponding to the wavelength C is recorded.

[0036] It should be noted that this process ensures that the system can capture important spectral information of the target object, providing a reliable basis for subsequent material identification, classification or sorting based on wavelength characteristics.

[0037] In another preferred embodiment of the present application, the method for determining the material of the bottle to be sorted based on the wavelength C comprises: A confidence level z is preset, the wavelength Z corresponding to the PET material absorption peak is obtained, and a wavelength floating interval [Z-Z sta / z, Z+Z sta / z] is set, and if the wavelength C overlaps with the wavelength floating interval by more than 90%, the material of the bottle to be sorted is determined to be PET.

[0038] It should be noted that the chemical bonds in plastic bottles selectively absorb certain wavelengths of light, while allowing other wavelengths of light to pass through or be reflected, depending on their own properties.

[0039] In another preferred embodiment of the present invention, the maximum imaging time T is obtained. max Ensure distance L > T max ×v.

[0040] Understandably, during the initial screening, it is necessary to ensure that the conveyor belt is long enough to prevent the plastic bottles from moving to the second step before the initial screening is completed.

[0041] In a preferred embodiment, if the overlap between wavelength C and other wavelength floating ranges (excluding PET materials) exceeds 50%, subsequent operations are stopped, and the corresponding plastic bottles are re-sorted.

[0042] It is worth noting that for ambiguous plastic bottles, if they cannot be distinguished at present, they need to be re-sorted to avoid sorting errors.

[0043] A vision-based plastic bottle sorting and selection vision control system includes: Imaging module: When a plastic bottle is detected on the conveyor belt, the camera is activated to acquire the image height H of the plastic bottle. The image height H of the plastic bottle represents the number of pixels in a single row that the camera captures and stitches together. The imaging time T is calculated as T = H / FL, where FL represents the line frequency of the camera. Preliminary screening module: Perform the following operations on the image in sequence: Obtain the exposure value E of each pixel in the image, and preset the exposure range [E]. min E max Pixels whose exposure value E is outside the exposure range are marked as abnormal pixels, where E min E max These represent the preset minimum and maximum exposure values, respectively; Calculate the proportion of abnormal pixels B = N1 / N all N1 represents the number of abnormal pixels, N all The number of pixels occupied by plastic bottles in the image is represented by the proportion B of abnormal pixels, which is used to initially screen plastic bottles. Material identification module: Plastic bottles that pass the initial screening are marked as bottles to be sorted. An NIR sensor is set at a preset distance L meters from the camera. The NIR sensor includes a transmitter and a receiver. When the bottle to be sorted passes directly below the transmitter, it emits near-infrared light and acquires the spectrum reflected by the bottle. Based on the spectrum, the wavelength C of the corresponding absorption peak of the bottle to be sorted is determined. The sorting module determines the material of the bottle to be sorted based on the wavelength C, obtains the time t1 when the near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined after the material confirmation is completed, and calculates the total time T all =T+t2-t1, and calculates the distance coefficient S=L+T all S×v, where v represents the speed of the conveying belt, and the air valve corresponding to the distance coefficient S is started to blow out the plastic bottle with the material PET and complete the sorting.

[0044] The above describes one embodiment of the present application in detail, but the content is only the preferred embodiment of the present application and cannot be considered to limit the implementation range of the present application. Any equivalent changes and improvements made within the scope of the present application should still belong to the scope of the present application.

Claims

1. A vision-based control method for sorting and selecting plastic bottles, characterized in that, Includes the following steps: S1: When a plastic bottle is detected on the conveyor belt, the camera is activated to acquire the image height H of the plastic bottle. The image height H of the plastic bottle represents the number of pixels in a single row that the camera captures and stitches together. The imaging time T is calculated as T = H / FL, where FL represents the line frequency of the camera. S2: Perform the following operations on the image sequentially: Obtain the exposure value E of each pixel in the image, and preset the exposure range [E]. min E max Pixels whose exposure value E is outside the exposure range are marked as abnormal pixels, where E min E max These represent the preset minimum and maximum exposure values, respectively; Calculate the proportion of abnormal pixels B = N1 / N all N1 represents the number of abnormal pixels, N all The number of pixels occupied by plastic bottles in the image is represented by the proportion B of abnormal pixels, which is used to initially screen plastic bottles. S3: Record the plastic bottles that have passed the initial screening as bottles to be sorted. Set up an NIR sensor at a preset distance L meters from the camera. The NIR sensor includes a transmitter and a receiver. When the bottle to be sorted passes directly below the transmitter, it emits near-infrared light and obtains the spectrum reflected by the bottle. Based on the spectrum, determine the wavelength C of the corresponding absorption peak of the bottle to be sorted. S4: Determine the material of the bottle to be sorted based on wavelength C. After material confirmation, obtain the time t1 when near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined, and calculate the total time T. all =T+t2-t1, calculate the distance coefficient S=L+T all ×v, where v represents the conveyor belt speed, and the air valve corresponding to the distance coefficient S is activated to blow out plastic bottles made of PET and complete the sorting.

2. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 1, characterized in that, In step S1, the minimum exposure time T is preset. min If the line frequency FL ≤ 1 / T min The line frequency FL is increased by a preset gradient coefficient λ×f until the line frequency FL > 1 / T. min , where λ represents the preset gradient value, λ=1,2,...,f represents the preset minimum row frequency.

3. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 1, characterized in that, In step S2, the method for preliminary screening of plastic bottles based on the proportion B of abnormal pixels includes: Plastic bottles with abnormal pixel proportions B ≥ 10% are blown off the conveyor belt through an air valve, and the average exposure value E is calculated. ave The mean E ave >E max Plastic bottles that are overexposed are classified as overexposed plastic bottles, with a mean E. ave <E max Plastic bottles are classified as low-exposure plastic bottles; After one round of sorting is completed, the light intensity is reduced by 10% from the original light intensity, and the overexposed plastic bottles are put back into the conveyor belt for sorting. The above operation is repeated until there are no more overexposed plastic bottles. Increase the light intensity by 10% on top of the original intensity, and put the low-exposure plastic bottles back into the conveyor belt for sorting. Repeat the above operation until there are no more low-exposure plastic bottles.

4. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 1, characterized in that, In step S3, the method for determining the wavelength C of the absorption peak corresponding to the bottle to be sorted based on the spectrum includes: A preset judgment threshold is set. If the absolute value of the absorption peak exceeds the judgment threshold, the corresponding wavelength range is recorded as wavelength C.

5. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 1, characterized in that, In step S4, the method for determining the material of the bottle to be sorted based on wavelength C includes: Set a preset confidence level z, obtain the wavelength Z corresponding to the absorption peak of PET material, and set the wavelength floating range [ZZ]. sta / z,Z+Z sta If the overlap between wavelength C and the wavelength floating range exceeds 90%, then the material of the bottle to be sorted is determined to be PET.

6. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 1, characterized in that, In step S3, the maximum imaging time T is obtained. max Ensure distance L > T max ×v.

7. The visual control method for sorting and selecting plastic bottles based on visual detection according to claim 5, characterized in that, If wavelength C overlaps with more than 50% of the floating range of wavelengths other than PET materials, then the subsequent operation is stopped and the corresponding plastic bottle is re-sorted.

8. A vision control system for sorting and selecting plastic bottles based on vision detection, characterized in that, include: Imaging module: When a plastic bottle is detected on the conveyor belt, the camera is activated to acquire the image height H of the plastic bottle. The image height H of the plastic bottle represents the number of pixels in a single row that the camera captures and stitches together. The imaging time T is calculated as T = H / FL, where FL represents the line frequency of the camera. Preliminary screening module: Perform the following operations on the image in sequence: Obtain the exposure value E of each pixel in the image, and preset the exposure range [E]. min E max Pixels whose exposure value E is outside the exposure range are marked as abnormal pixels, where E min E max These represent the preset minimum and maximum exposure values, respectively; Calculate the proportion of abnormal pixels B = N1 / N all N1 represents the number of abnormal pixels, N all The number of pixels occupied by plastic bottles in the image is represented by the proportion B of abnormal pixels, which is used to initially screen plastic bottles. Material identification module: Plastic bottles that pass the initial screening are marked as bottles to be sorted. An NIR sensor is set at a preset distance L meters from the camera. The NIR sensor includes a transmitter and a receiver. When the bottle to be sorted passes directly below the transmitter, it emits near-infrared light and acquires the spectrum reflected by the bottle. Based on the spectrum, the wavelength C of the corresponding absorption peak of the bottle to be sorted is determined. Sorting module: Based on wavelength C, it determines the material of the bottle to be sorted. After material confirmation, it acquires the time t1 when near-infrared light is emitted and the time t2 when the material of the bottle to be sorted is determined, and calculates the total time T. all =T+t2-t1, calculate the distance coefficient S=L+T all ×v, where v represents the conveyor belt speed, and the air valve corresponding to the distance coefficient S is activated to blow out plastic bottles made of PET and complete the sorting.

Images