Machine vision-based visual bad food rejecting system

Abstract

The invention belongs to the technical field of visual rejecting of bad food, and particularly discloses a visual rejecting system for bad food based on machine vision. The system comprises a food information recording module, a food information monitoring module, a food qualification analysis module, an undesirable food confirmation module, an undesirable food monitoring module, an undesirable food removal module, a food packaging monitoring module, a food packaging analysis module and an undesirable product removal terminal. The food qualification analysis of each food is performed according to the appearance information and the oil stain information of each food, so that the guarantee comprehensiveness of the food quality is improved, the purchase willingness of consumers and the overall evaluation of the product quality are improved, and meanwhile, the packaging analysis of each product is performed according to the packaging information of each product after the food is packaged, so that the product quality is improved. The problem that the analysis attention degree of the packaging qualification index of the packaged product is insufficient at present is solved, and therefore the packaging integrity of the product is guaranteed.

Description

Technical Field

[0001] The invention belongs to the technical field of visual rejection of bad food, and relates to a visual rejection system of bad food based on machine vision. Background Art

[0002] Bad food usually refers to food with appearance defects, poor packaging, quality problems or containing foreign matter, while bad puffed food refers to puffed food that does not meet relevant quality standards, safety standards or has problems that affect consumers' health and experience. Therefore, it needs to be visually eliminated during the production process.

[0003] For example, the Chinese invention patent with publication number CN115222689A discloses a food safety visualization system based on AI visual recognition and blockchain technology, including: a visual recognition module, used to identify food types and obtain food information through AI visual recognition technology. A traceability module, connected to the visual recognition module, is used to trace the source of food through the blockchain network and obtain the quality inspection results of the food. A visualization module, connected to the visual recognition module, is used to present the food information and the quality inspection results of the food. A complete food safety traceability system is constructed by the three parts of image recognition, tracing, and visualization, which can complete the control of food safety information, track the source of food, and reduce the occurrence of food safety problems. It has the characteristics of simple and convenient operation, safe and reliable data sources, clear and concise presentation effects, and wide technical and commercial application value.

[0004] The above existing technologies have the following deficiencies: 1. Currently, the testing and analysis of food mainly focuses on the safety traceability of food, and insufficient attention is paid to the testing and analysis of qualified food, which makes it impossible to ensure the comprehensiveness of food quality assurance, thus affecting consumers' willingness to buy and their overall evaluation of product quality.

[0005] 2. Currently, food testing is mainly focused on testing food after production, that is, insufficient attention is paid to the analysis of the packaging qualification index of packaged products, which cannot guarantee the qualification of product packaging, and thus cannot guarantee the packaging integrity of the product, and further cannot guarantee the safety of the product. Summary of the invention

[0006] In view of this, in order to solve the problems raised in the above background technology, a defective food visual rejection system based on machine vision is proposed.

[0007] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a visual rejection system for bad food based on machine vision, including: a food information recording module for recording standard information of food, equipment information of food monitoring equipment and standard information of products.

[0008] The food information monitoring module is used to monitor the appearance information and oil stain information of each food. The appearance information includes color information, surface information and shape information. The oil stain information includes the number of oil stains, the area of ​​each oil stain and the color chromaticity value.

[0009] The food qualification analysis module is used to perform food qualification analysis on each food according to the appearance information and oil pollution information of each food, and obtain the food qualification index of each food.

[0010] The bad food confirmation module is used to confirm each bad food according to the food qualification index of each food.

[0011] The bad food monitoring module is used to monitor the parameter information of each bad food, and the parameter information includes the current position coordinates, the current monitoring time point and the current running speed.

[0012] The bad food rejection module is used to confirm the rejection information of each bad food according to the parameter information of each bad food, and execute the rejection of the bad food.

[0013] The food packaging monitoring module is used to package the rejected food after the bad food is rejected, and to monitor the packaging information of each product after the food is packaged.

[0014] The food packaging analysis module is used to perform packaging analysis on each product according to the packaging information of each product after food packaging, and obtain the packaging qualification index of each product.

[0015] The defective product rejection terminal is used to identify each defective product according to the packaging qualification index of each product, and then execute the rejection of the defective products.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention avoids the shortcomings of the current manual rejection of bad foods by confirming the rejection information of each bad food according to the parameter information of each bad food and executing the rejection of the bad food, thereby improving the efficiency of food rejection and further ensuring the accuracy of bad food identification, thereby improving food production efficiency. At the same time, the bad food is rejected by machine, which reduces the cost of rejecting bad food.

[0017] (2) The present invention analyzes the quality of each food based on its appearance information and oil pollution information, thus breaking the current lack of attention paid to the detection and analysis of food quality, thereby improving the comprehensiveness of food quality assurance, thereby increasing consumers' willingness to buy and their overall evaluation of product quality.

[0018] (3) The present invention solves the current problem of insufficient attention paid to the analysis of the packaging qualification index of packaged products by performing packaging analysis on each product according to the packaging information of each product after food packaging, thereby improving the qualification of product packaging, thereby ensuring the packaging integrity of the product and further improving the safety of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0020] Figure 1 It is a schematic diagram of the connection of each module of the system of the present invention.

[0021] Figure 2 It is a schematic diagram of the connection of each bad food confirmation step of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] See also Figure 1 As shown, the present invention provides a defective food visual rejection system based on machine vision, the system comprising: a food information recording module, a food information monitoring module, a food qualification analysis module, a defective food confirmation module, a defective food monitoring module, a defective food rejection module, a food packaging monitoring module, a food packaging analysis module and a defective product rejection terminal.

[0024] In the above, the food qualification analysis module is respectively connected to the food information recording module, the food information monitoring module and the bad food confirmation module, the bad food confirmation module is respectively connected to the food information recording module and the bad food monitoring module, the bad food rejection module is respectively connected to the food information recording module, the bad food monitoring module and the food packaging monitoring module, and the food packaging analysis module is also respectively connected to the food information recording module, the food packaging monitoring module and the bad product rejection terminal.

[0025] The food information recording module is used to record the standard information of food, the equipment information of food monitoring equipment and the standard information of products.

[0026] It should be added that the standard information of food includes: standard length of food, standard sealing width, standard thickness, standard color chromaticity value and standard food qualification index; the equipment information of food monitoring equipment includes: the position coordinates of the rejection point; the standard information of the product includes: the allowable test gas content difference of the product, the allowable wrinkle height of the sealing surface, the standard sealing width and the standard packaging qualification index.

[0027] The food information monitoring module is used to monitor the appearance information and oil stain information of each food. The appearance information includes color information, surface information and shape information. The oil stain information includes the number of oil stains, the area of ​​each oil stain and the color chromaticity value.

[0028] It should be supplemented that the color information of each food includes: the number of colors of each food and the chromaticity value of each color, the appearance information of each food includes: the number of damaged parts of each food, the area of ​​each damaged part, the number of scratches and the length of each scratch, and the shape information of each food includes: the length, width and thickness of each food.

[0029] It should be added that the process of obtaining the number of colors of each food and the chromaticity of each color is as follows: the image of the food is captured by a camera, and then the image is analyzed by using professional image analysis software, such as Adobe Photoshop, ImageJ, etc., so as to calculate the chromaticity value of each color in each food, and at the same time count the number of colors of each food.

[0030] It should be added that the area of ​​each damaged part of each food is obtained by capturing images of the food through a camera. The image acquisition card converts the analog image signal obtained by the camera into a digital signal and transmits it to a computer. The image is then analyzed by the image processing software in the computer. For detecting damaged parts of food, the image processing software determines the damaged area by identifying changes in features such as texture, color and shape of the food surface in the image. For example, the surface texture of an intact food is continuous and the color is uniform, while the damaged parts will have texture interruptions and color differences. Therefore, by counting the damaged parts of each food, the number of damaged parts of each food is obtained.

[0031] It should be added that the number and length of scratches on each food are obtained by capturing images of the food with a camera. The image acquisition card converts the analog image signals obtained by the camera into digital signals and transmits them to the computer. The scratches are then detected through an image processing algorithm. For example, scratches are identified by analyzing the changes in grayscale values ​​in the image. The grayscale values ​​of the scratched area are often different from those of the surrounding normal areas. The algorithm can determine the boundaries of the scratches based on these differences, thereby calculating the number and length of the scratches.

[0032] It should be added that the number of oil stains, the area of ​​each oil stain and the color chromaticity value are obtained by analyzing the collected food images using image processing algorithms. There are differences between the oil stain area and the surrounding normal area in color, grayscale value, texture, etc. The oil stain area can be segmented from the image by a set threshold, thereby determining the number of oil stains and calculating its area. The color chromaticity value of the oil stain is obtained by color analysis of the segmented oil stain area (for example, analysis in RGB color space or other color space).

[0033] The food qualification analysis module is used to perform food qualification analysis on each food according to the appearance information and oil pollution information of each food, and obtain the food qualification index of each food.

[0034] Exemplarily, the food quality analysis of each food includes: A1, according to the appearance information of each food, calculating the appearance quality index δ of each food i , i is the food number, i=1,2,......,m.

[0035] Furthermore, the statistical analysis of the appearance qualification index of each food includes: A1-1, according to the color information of each food, statistical analysis of the color qualification index of each food

[0036] Furthermore, the statistical color qualification index of each food includes: A1-1-1, extracting the number of colors of each food and the chromaticity value of each color from the color information of each food, and recording them as D i and R ij , j is the color number, j=1,2,......,n.

[0037] A1-1-2. Extract the standard color chromaticity value of the food from the standard information of the food and record it as R′.

[0038] A1-1-3. Calculate the chromaticity difference Δα of each color of each food ij , Δα ij =|R ij -R′|, and then from Δα ij The maximum value is selected as the maximum color chromaticity difference of each food and recorded as Δα′ i .

[0039] A1-1-4. Δα ij Calculate the mean value and get the average color chromaticity difference of each food, recorded as

[0040] A1-1-5. Calculate the color qualification index of each food D0 and α′ are the number of reference colors and color chromaticity difference respectively, and e is a natural constant.

[0041] It should be added that D0 and α′ are both extracted from the technical specification table for visual elimination of defective food.

[0042] A1-2. Extract the standard length, standard sealing width and standard thickness of the food from the standard information of the food.

[0043] A1-3. Extract the length, width and thickness of each food from the shape information of each food, and subtract them from the standard length, standard sealing width and standard thickness respectively to obtain the length difference, width difference and thickness difference of each food, and record them as L respectively. i , W i and H i .

[0044] A1-4. Calculate the shape qualification index of each food L', W' and H' are the set allowable length difference, allowable width difference and allowable thickness difference respectively.

[0045] It should be added that L′, W′ and H′ are all extracted from the technical specification table for visual rejection of defective food.

[0046] A1-5. Extract the number of damaged parts and the area of ​​each damaged part, as well as the number of scratches and the length of each scratch from the appearance information of each food, and then calculate the appearance qualification index of each food

[0047] Furthermore, the statistical analysis of the apparent qualified index of each food includes: A1-5-1, summing up the areas of the damaged parts of each food to obtain the total area of ​​the damaged parts of each food, and recording it as S i .

[0048] A1-5-2. Record the number of damaged parts of each food as D′ i , and then calculate the damage qualified index λ of each food i , D′0 and S′ are the set allowable number of damaged areas and allowable area of ​​damaged areas respectively.

[0049] It should be added that D′0 and S′ are both extracted from the technical specification table for visual rejection of defective food.

[0050] A1-5-3. From the scratch lengths of each food, select the maximum scratch length of each food and record it as l i .

[0051] A1-5-4. Calculate the mean of the scratch lengths of each food to obtain the average scratch length of each food, recorded as The number of scratches on each food is recorded as D″ i .

[0052] A1-5-5. Calculate the scratch pass index λ′ of each food i , D″0 and l′ are the set permissible number of scratches and permissible scratch length respectively.

[0053] A1-5-6, from λ i and λ′ i The minimum value is selected as the apparent qualified index of each food and recorded as

[0054] A1-6. Calculate the appearance qualification index of each food i ,

[0055] A2. According to the oil pollution information of each food, calculate the oil pollution qualified index φ of each food i .

[0056] Furthermore, the statistical oil stain qualified index of each food includes: A2-1, matching the color chromaticity value of each oil stain in each food with the oil stain color chromaticity value interval corresponding to each color influencing factor, obtaining the color chromaticity value influencing factor of each oil stain in each food, and recording it as θ ip , p is the oil pollution number, p=1,2,......,q.

[0057] A2-2. Extract the length and width of each food, and then calculate the area of ​​each food, denoted as S″′ i .

[0058] A2-3. The area of ​​each oil stain in each food is recorded as S ip , and then calculate the oil pollution impact ratio K of each food i ,

[0059] A2-4. Record the number of oil stains on each food as D″′ i , and then calculate the oil pollution qualified index φ of each food i , β′ and D″′0 are the oil pollution influence ratio and the number of oil pollution of the set reference respectively.

[0060] It should be added that β′ and D″′0 are both extracted from the technical specification table for visual elimination of defective food.

[0061] A3. Calculate the food quality index of each food r1 and r2 are the weights of the set appearance qualification index and oil pollution qualification index respectively, r1+r2=1, r1>r2.

[0062] It should be added that the appearance qualification index can comprehensively reflect the quality control conditions of multiple links in the food production process. For example, whether the shape is regular or not is related to the precision and stability of the production equipment, and the color uniformity is closely related to the quality of the raw materials and the consistency of the processing technology. Oil pollution will affect the appearance and hygiene of the food, but as long as the degree of oil pollution is not serious, the impact on the taste and nutritional value of the food is relatively small. For example, a small amount of surface oil pollution may not have a significant effect on the crispy taste of puffed food. Therefore, r1>r2 is set. For the convenience of analysis, r1 can be specifically taken as 0.6 and r2 can be specifically taken as 0.4.

[0063] The embodiments of the present invention perform food qualification analysis on each food based on its appearance information and oil pollution information, thereby breaking the current lack of attention paid to the detection and analysis of food qualification, and further improving the comprehensiveness of food quality assurance, thereby increasing consumers' willingness to buy and their overall evaluation of product quality.

[0064] The bad food confirmation module is used to confirm each bad food according to the food qualification index of each food.

[0065] See also Figure 2 As shown, illustratively, the confirmation of each bad food includes: Q1, extracting the food qualification index of each food.

[0066] Q2. Extract the standard food conformity index of food from the standard information of food.

[0067] Q3. Compare the food qualification index of each food with the standard food qualification index of the food.

[0068] Q4. If the qualified index of a food is less than the standard food qualified index of the food, the food will be recorded as a bad food, and then the bad foods will be obtained.

[0069] The bad food monitoring module is used to monitor parameter information of each bad food, and the parameter information includes current position coordinates, current monitoring time point and current running speed.

[0070] It should be added that the process of obtaining the current position coordinates of each defective food is as follows: the camera takes a picture of the inside of the food monitoring equipment, obtains the image containing the food and transmits it to the computer through the image acquisition card. The image processing algorithm in the computer analyzes the shape, color, texture and other characteristics of the food and compares them with the reference points or areas inside the equipment to determine the position of the defective food in the equipment. For example, the coordinate range of different areas inside the equipment can be pre-set, and the pixel position of the defective food in the image can be identified and converted into the actual current position coordinates.

[0071] It should be added that the current monitoring time point of each bad food is obtained by monitoring the clock module inside the bad food rejection device.

[0072] It should be added that the current running speed of each bad food is obtained by emitting a laser beam to the surface of the bad food through a laser speed meter. When the food moves, the frequency of the reflected light will change according to the Doppler effect. By measuring the change in the reflected light frequency Δf, the wavelength λ of the laser is known, and according to the formula (where v is the running speed of the bad food), calculate the current running speed of each bad food.

[0073] The bad food rejection module is used to confirm the rejection information of each bad food according to the parameter information of each bad food, and execute the rejection of the bad food.

[0074] The embodiment of the present invention confirms the rejection information of each defective food according to the parameter information of each defective food and executes the rejection of the defective food, thereby avoiding the shortcomings of the current manual rejection of defective food, improving the efficiency of food rejection, and further ensuring the accuracy of defective food identification, thereby improving food production efficiency. At the same time, the defective food is rejected by machine, which reduces the cost of rejecting defective food.

[0075] Exemplarily, the confirmation of rejection information of each bad food includes: marking the current position coordinates of each bad food as (x i ,y i ).

[0076] The position coordinates of the rejection point are extracted from the equipment information of the food monitoring equipment and recorded as (x′, y′).

[0077] Count the distance d between each bad food and the rejection point i ,

[0078] In a specific embodiment, the coordinates of the bad food are detected as (301.5 mm, 149.8 mm), and the coordinates of the rejection point are (1799.2 mm, 400.5 mm).

[0079] Count the distance d between bad food and the rejection point,

[0080] The current running speed of each bad food is recorded as v i .

[0081] Count the time t for each bad food to reach the rejection point i ,

[0082] The current monitoring time point of each defective food is extracted, and then according to the time it takes for each defective food to reach the rejection point, the rejection monitoring time point of each defective food reaching the rejection point is obtained, and it is used as the rejection information of each defective food.

[0083] The food packaging monitoring module is used to package the rejected food after the bad food is rejected, and monitor the packaging information of each product after the food is packaged.

[0084] The food packaging analysis module is used to perform packaging analysis on each product according to the packaging information of each product after food packaging, and obtain the packaging qualification index of each product.

[0085] Exemplarily, the packaging analysis of each product includes: extracting the gas content of each product before and after the gas content test from the packaging information of each product after food packaging, and recording them as ξ u and ξ′ u , u is the product number, u=1,2,......,b.

[0086] The permissible test gas content difference of the product is extracted from the standard information of the product and recorded as Δξ.

[0087] Statistical gas content qualification index θ of each product u ,

[0088] The height of each wrinkle on the sealing surface of each product is extracted from the packaging information of each product after food packaging, and then the maximum wrinkle height on the sealing surface of each product is extracted, which is recorded as h u .

[0089] The permissible wrinkle height on the sealing surface of the product is extracted from the standard information of the product and recorded as h′.

[0090] If the maximum wrinkle height of a product's sealing surface is less than the permissible wrinkle height of the sealing surface, the sealing flatness qualified index of the product is recorded as 1, otherwise, the sealing flatness qualified index of the product is recorded as 0. In this way, the sealing flatness qualified index of each product is obtained and recorded as θ′ u ,θ′ u The value is 1 or 0.

[0091] Extract the sealing width of each product from the packaging information of each product after food packaging, and record it as ω u , and extract the standard sealing width of the product from the standard information of the product, recorded as ω′, and then according to θ u The statistical method is similar to that of the sealing width of each product, and it is recorded as θ″ u .

[0092] Statistics of packaging qualification index of each productu , l u =θ u *z1+θ′ u *z2+θ″ u *z3, z1, z2 and z3 are the weights of the set gas content qualified index, sealing flatness qualified index and sealing width qualified index respectively, z1+z2+z3=1, z1>z2>z3.

[0093] It should be added that the gas content qualified index is a comprehensive consideration of the overall sealing of the package. Whether it is the tiny gap at the seal, the micropores of the packaging material itself, or the possible leakage points in other parts of the package, they will affect the gas content. The poor flatness of the seal, such as wrinkles or bulges, will affect the sealing effect of the seal to a certain extent, but this effect may not be as direct and serious as gas leakage. Although the seal width has a certain impact on the sealing effect of the seal, as long as the seal width is within a reasonable deviation range, the seal can still achieve basic sealing functions. For example, the seal width is slightly narrower or wider, as long as it does not exceed the critical value that can ensure sealing, it will not cause fundamental damage to the barrier effect of the gas in the package. Therefore, z1>z2>z3 is set. For the convenience of analysis, z1 can be specifically taken as 0.5, z2 can be specifically taken as 0.3, and z3 can be specifically taken as 0.2.

[0094] The defective product rejection terminal is used to identify each defective product according to the packaging qualification index of each product, and then execute the rejection of the defective products.

[0095] Exemplarily, the confirming each defective product includes: extracting a standard packaging qualification index of the product from standard information of the product.

[0096] Compare the packaging qualification index of each product with the standard packaging qualification index of the product.

[0097] If the package qualification index of a product is less than the standard package qualification index of the product, the product is recorded as a defective product, thereby obtaining each defective product.

[0098] The embodiment of the present invention solves the current problem of insufficient attention paid to the analysis of the packaging qualification index of the packaged products by performing packaging analysis on each product according to the packaging information of each product after food packaging, improves the qualification of product packaging, thereby ensuring the packaging integrity of the product and further improving the safety of the product.

[0099] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A visual rejection system for bad food based on machine vision, characterized by: The system includes: Food information recording module, used to record food standard information, equipment information of food monitoring equipment and product standard information; The food information monitoring module is used to monitor the appearance information and oil stain information of each food. The appearance information includes color information, surface information and shape information. The oil stain information includes the number of oil stains, the area of ​​each oil stain and the color chromaticity value. A food qualification analysis module is used to perform food qualification analysis on each food according to the appearance information and oil pollution information of each food, and obtain the food qualification index of each food; The bad food confirmation module is used to confirm each bad food according to the food qualification index of each food; The bad food monitoring module is used to monitor the parameter information of each bad food, and the parameter information includes the current position coordinates, the current monitoring time point and the current running speed; The bad food rejection module is used to confirm the rejection information of each bad food according to the parameter information of each bad food, and execute the rejection of the bad food; The food packaging monitoring module is used to package the rejected food after the bad food is rejected, and monitor the packaging information of each product after the food is packaged; The food packaging analysis module is used to perform packaging analysis on each product according to the packaging information of each product after food packaging, and obtain the packaging qualification index of each product; The defective product rejection terminal is used to identify each defective product according to the packaging qualification index of each product, and then execute the rejection of the defective products.

2. According to claim 1, a visual rejection system for bad food based on machine vision is characterized by: The food quality analysis of each food includes: A1. According to the appearance information of each food, calculate the appearance qualification index of each food δ i , i is the food number, i=1,2,......,m; A2. According to the oil pollution information of each food, calculate the oil pollution qualified index φ of each food i ; A3. Calculate the food quality index of each food r1 and r2 are the weights of the set appearance qualification index and oil pollution qualification index respectively, r1+r2=1, r1>r2.

3. The system for visually removing bad food based on machine vision according to claim 2, characterized in that: The statistical appearance qualification index of each food includes: According to the color information of each food, the color qualification index of each food is calculated Extracting the standard length, standard sealing width and standard thickness of the food from the standard information of the food; The length, width and thickness of each food are extracted from the shape information of each food, and the length, width and thickness of each food are subtracted from the standard length, standard sealing width and standard thickness to obtain the length difference, width difference and thickness difference of each food, which are recorded as L i , W i and H i ; Calculate the shape qualification index of each food L′, W′ and H′ are the set permissible length difference, permissible width difference and permissible thickness difference respectively; Extract the number of damaged parts and the area of ​​each damaged part, the number of scratches and the length of each scratch from the surface information of each food, and then calculate the surface qualification index of each food Calculate the appearance qualification index of each food δ i , 4. The machine vision-based visual rejection system for bad food according to claim 2, characterized in that: The statistical color qualification index of each food includes: The number of colors and the chromaticity value of each color are extracted from the color information of each food and recorded as D i and R ij , j is the color number, j=1,2,......,n; Extract the standard color chromaticity value of the food from the standard information of the food and record it as R′; Statistical analysis of the chromaticity difference Δα of each color of each food ij , Δα ij =|R ij -R′|, and then from Δα ij The maximum value is selected as the maximum color chromaticity difference of each food and recorded as Δα i ′; Δα ij Calculate the mean value and get the average color chromaticity difference of each food, recorded as Statistical analysis of the color qualification index of each food D0 and α′ are the number of reference colors and color chromaticity difference respectively, and e is a natural constant.

5. The system for visually removing bad food based on machine vision according to claim 4 is characterized in that: The statistical apparent qualified index of each food includes: The total area of ​​the damaged parts of each food is obtained by summing up the damaged areas of each food and recorded as S i ; The number of damaged parts of each food is recorded as D i ′, and then calculate the damage qualified index λ of each food i , D0′ and S′ are the number of permissible damage and the permissible damage area respectively; The maximum scratch length of each food is selected from the scratch lengths of each food and recorded as l i ; The average scratch length of each food is calculated to obtain the average scratch length of each food, which is recorded as The number of scratches on each food is recorded as D i ″; Calculate the scratch pass index λ of each food i ′, D0″ and l′ are the set permissible number of scratches and permissible scratch length, respectively; From λ i and λ i ′, the minimum value is selected as the apparent qualified index of each food and recorded as 6. The machine vision-based visual rejection system for bad food according to claim 2, characterized in that: The above mentioned statistics on the oil pollution qualified index of each food include: Match the color chromaticity value of each oil stain in each food with the oil stain color chromaticity value interval corresponding to each color influencing factor to obtain the color chromaticity value influencing factor of each oil stain in each food, and record it as θ ip , p is the oil pollution number, p=1,2,......,q; Extract the length and width of each food, and then calculate the area of ​​each food, denoted as S i ″′; The area of ​​each oil stain in each food is recorded as S ip , and then calculate the oil pollution impact ratio K of each food i , The number of oil stains on each food is recorded as D i ″′, and then calculate the oil pollution qualified index of each food β′ and D0″′ are the oil pollution influence ratio and the number of oil pollution of the set reference respectively.

7. The machine vision-based visual rejection system for bad food according to claim 2, characterized in that: The identification of each undesirable food includes: Q1. Extract the food qualification index of each food; Q2. Extract the food standard food qualification index from the food standard information; Q3. Compare the food qualification index of each food with the standard food qualification index of the food; Q4. If the qualified index of a food is less than the standard food qualified index of the food, the food will be recorded as a bad food, and then the bad foods will be obtained.

8. The machine vision-based visual rejection system for bad food according to claim 1, characterized in that: The confirmation of rejection information of each bad food includes: The current position coordinates of each unhealthy food are marked as (x i ,y i ); Extract the position coordinates of the rejection point from the equipment information of the food monitoring equipment and record them as (x′, y′); Count the distance d between each bad food and the rejection point i , The current running speed of each bad food is recorded as v i ; Count the time t for each bad food to reach the rejection point i , The current monitoring time point of each defective food is extracted, and then according to the time it takes for each defective food to reach the rejection point, the rejection monitoring time point of each defective food reaching the rejection point is obtained, and it is used as the rejection information of each defective food.

9. The machine vision-based visual rejection system for bad food according to claim 1, characterized in that: The packaging analysis of each product includes: The gas content of each product before and after the gas content test is extracted from the packaging information of each product after food packaging, and recorded as ξ u and u ′, u is the product number, u=1,2,......,b; Extract the permitted test gas content difference of the product from the standard information of the product and record it as Δξ; Statistical gas content qualification index θ of each product u , The height of each wrinkle on the sealing surface of each product is extracted from the packaging information of each product after food packaging, and then the maximum wrinkle height on the sealing surface of each product is extracted, which is recorded as h u ; Extract the permissible wrinkle height of the product sealing surface from the product standard information, denoted as h′; If the maximum wrinkle height of a product's sealing surface is less than the permissible wrinkle height of the sealing surface, the sealing flatness qualified index of the product is recorded as 1, otherwise, the sealing flatness qualified index of the product is recorded as 0. In this way, the sealing flatness qualified index of each product is obtained and recorded as θ′ u ,θ′ u The value is 1 or 0; Extract the sealing width of each product from the packaging information of each product after food packaging, and record it as ω u , and extract the standard sealing width of the product from the standard information of the product, recorded as ω′, and then according to θ u The statistical method is similar to that of the sealing width qualification index of each product, and it is recorded as θ″ u ; Statistics of packaging qualification index of each product u , l u =θ u *z1+θ′ u *z2+θ″ u *z3, z1, z2 and z3 are the weights of the set gas content qualified index, sealing flatness qualified index and sealing width qualified index respectively, z1+z2+z3=1, z1>z2>z3.

10. The machine vision-based visual rejection system for bad food according to claim 9, characterized in that: The confirmation of each defective product includes: Extract the standard packaging qualification index of the product from the standard information of the product; Compare the packaging qualification index of each product with the standard packaging qualification index of the product; If the package qualification index of a product is less than the standard package qualification index of the product, the product is recorded as a defective product, thereby obtaining each defective product.

Images