Apparatus and method based on the online detection of soft capsule quality of computer vision
Technical field
The present invention relates to a kind of characteristic and carry out machine vision detection device and the method that real-time online detects, refer in particular to a kind of apparatus and method of the online detection of soft capsule quality based on computer vision according to article.
Background technology
Soft capsule is a kind of novel form that grows up behind sheet, injection, can and be encapsulated in the glued membrane the quantitative pressure injection of oily medicine, drug solution or drug suspension, pastel even drug powder, forms big or small, different seal capsule.Compare with other formulations, possess bioavilability height, content accurately, characteristics such as good uniformity, good looking appearance: if the oily medicine, also can save technical finesses such as absorption, curing, can effectively avoid the oily medicine from absorb auxiliary material, to ooze out, so soft capsule is the optimum formulation of oiliness medicine.In addition, the hydrophobic drug of low melting point substance, bioavilability difference, bad bitter taste and the medicine smelling the medicine, trace active medicine of flavor and meet light, wet, thermally labile and easy oxidation also are fit to make soft capsule.Soft capsule is at the still just popular gradually in recent years novel form of China, the material of various ways such as that but it is applicable to is aqueous, paste, powdery, handsome in appearance, be easy to take, drug effect is fast, these many advantages make soft capsule be applied to pharmacy industry more and more widely, and are subjected to the welcome of a plurality of industries such as health products, food, cosmetics very soon.The characteristic of soft capsule and production process have proposed very high requirement to equipment and technology, and the high-quality equipment of sophistication that will lean on of soft capsule guarantees, so just can satisfy the demands of consumers, and withstands intense market competition.At present, in order to guarantee the quality of soft capsule, manufacturer generally all adopts the mode of artificial manual inspection.This detection mode not only labour intensity height, labour cost is big, and because principal commander's property of manual detection can not guarantee the accurate detection to all capsule qualities.In addition, more existing capsule detection techniques are as correlative study patent " capsule automatic on-line weigh detection machine " (application number: 200420064679.2) because it is based on a kind of detection of weighing principle, only be suitable for the online detection of hard shell capsules.And this method can't be rejected the irregular soft capsule of those profiles (special-shaped ball).Numerous soft capsule factory commercial city proposes the requirement that realizes that soft capsule detects and rejects automatically by a grain quality (particularly weight).
Summary of the invention
For overcoming above-mentioned the deficiencies in the prior art, the invention provides a kind of method and apparatus that detects classification based on the online soft capsule quality of computer vision (particularly weight), can obtain the capsule image of those high-speed motions on carrier chain accurate and effective, all sidedly, handle by image, characteristic parameters such as the area of extraction capsule, size, shape are rejected those special-shaped capsules automatically according to these characteristic parameters on the one hand; With the weight opening relationships model of these characteristic parameters and capsule, judge rapidly whether the weight of this capsule is qualified on the other hand, and underproof capsule is rejected from production line automatically.
Technical scheme of the present invention is as follows:
Checkout gear of the present invention is arranged transport portion, Computer Vision Recognition part and capsule rejecting action part by capsule and is formed.The camera of Computer Vision Recognition part is installed in the carrier chain top that capsule is arranged transport portion, and capsule is rejected action part and is installed in the carrier chain side that capsule is arranged transport portion.It is characterized in that:
The function that described capsule is arranged the transport portion realization is to make the online orderly arrangement of soft capsule, main involving vibrations feed appliance and carrier chain.Because the transport portion of existing capsule checkout equipment all at the hard shell capsules design, directly applies to soft capsule with them and is easy to make the soft capsule breakage.The present invention is by on the support of transport portion front end vibration feeder being installed, make soft capsule become single-row order output, and the conduit that passes through portion mounted thereto is connected with carrier chain, in the pit that drops on carrier chain with one one of soft capsule, be fed forward with certain intervening sequences level with carrier chain, one one present to the computer vision part.
The function that described capsule is rejected the action part realization is to reject defective, the special-shaped soft capsules of those weight, mainly is linked in sequence by source of the gas, magnetic valve and nozzle and forms.Nozzle is installed in the side of carrier chain rear end, requires them to produce action according to the graded signal that computer vision partly provides, and the soft capsule that weight is defective, special-shaped is rejected from carrier chain accurately.
Described Computer Vision Recognition part, it is core of the present invention, is triggered sensor, control module and corresponding software and is formed by a camera system of lining by line scan, lighting box, a photoelectricity.The system of taking the photograph of wherein lining by line scan is linked in sequence and is formed by the camera of lining by line scan, a high speed image acquisition board, a computer.Photoelectricity triggers sensor and is installed on the carrier chain, links to each other with control module with high speed image acquisition board in the computer by circuit, and the triggering signal of image taking and the position of each capsule are provided; Lighting box is installed in the top of carrier chain, and the inside is mounted with the led light source of bar shaped, produces in capsule surface to diffuse uniformly.The camera of lining by line scan is installed in the lighting box, and perpendicular to be installed in carrier chain directly over, can photograph the image of capsule, the camera via line links to each other with high-speed figure image pick-up card in the computer, deposits captured image in computer.Computer links to each other with control module by data wire, and control module is sent the magnetic valve action that the signal controlling capsule is rejected action part, and the soft capsule that weight is defective, special-shaped is rejected from carrier chain accurately.
Detection method of the present invention is characterized in that:
(1). capsule is arranged transport portion makes soft capsule become single-row speed with 10~15/second to carry at a high speed with carrier chain forward automatically, guarantee by Computer Vision Recognition part time capsule can by present to the computer identification division;
(2). Computer Vision Recognition part is gathered image by being installed in the camera system of lining by line scan directly over the carrier chain to each capsules;
(3). collection capsule image is carried out sequence image handle, finish the extraction of characteristic parameters such as capsule area, shape, size, judge whether to special-shaped ball and determine whether this capsules weight is qualified by Intelligent Recognition software.
(4) for those special-shaped balls and the underproof capsule of weight, computer vision is partly sent instruction and is given control module, and control module is sent the magnetic valve action that the signal controlling capsule is rejected action part, and it is rejected from carrier chain accurately.
Described sequence image comprises each width of cloth Image Acquisition in handling, goes background, filtering, information storage etc.
Comprise some high-precision Intelligent Recognition models in the described Intelligent Recognition software, the capsule characteristics of image parameter that foundation is extracted and the relational model of special-shaped ball, and the relational model of capsule characteristics of image parameter and capsules weight, and guarantee that the Model Identification precision is more than 97%.
Light source in the lighting box can be selected and the position adjustment at different soft capsule objects, choose suitable light source frequency spectrum and locus, and make illumination uniformity in the visual field of camera, the position of camera in light-source box also can be adjusted according to the capsule object, makes camera can obtain the capsule image.
During work, soft capsule is arranged at capsule and is arranged in order on the transport portion and along with carrier chain advances, in the Computer Vision Recognition part, touch photoelectricity and trigger sensor, photoelectricity triggers sensor the capsule position is sent to control module and computer, and the high speed image acquisition board in the triggering computer, the image of collection capsule.Computer is in time handled captured sequence image after having taken at every turn, extracts characteristic parameters such as area, size, shape, and whether finish capsule by identification software is the detection of special-shaped ball and weight.To reject instruction by control module at last and send to capsule rejecting action part, produce action, special-shaped ball and the underproof capsule of weight will be rejected from carrier chain accurately by magnetic valve.Qualified soft capsule enters in the pulp-collecting box automatically.
Effect of the present invention is: (1) detection to soft capsule weight of the present invention is not the method for direct weighing, but characteristics of image parameter by setting up capsule and the relation between the capsules weight, reach the purpose of indirect detection capsules weight by computer vision, guarantee the accuracy of the online detection soft capsule of high speed weight.(2) the present invention also can finish the detection of special-shaped ball when finishing the soft capsule weight detecting.(3) the present invention guarantees that by the image information of the process software processing capsule of high speed the capsule detection speed is 10~15/second, just by work in one day 8 hours, and can go-on-go 28.8~43.2 ten thousand soft capsule.
Description of drawings
Fig. 1 soft capsule separator of the present invention schematic diagram;
Fig. 2 carrier chain enlarged drawing;
Fig. 3 light box part schematic diagram
The soft capsule weight of Fig. 4 computer vision prediction and the dependency relation figure of actual weight;
The 1-feed bin, the 2-vibration feeder, the 3-carrier chain, 4-photoelectricity triggers sensor, the 5-camera of lining by line scan, 6-air-blast atomizer, 7-pulp-collecting box, the 8-computer, 9-switch board, 10-carrier chain adjusting device, the 11-buncher, 12-carrier chain support, 13-high speed miniature electromagnetic valve, the clean high-pressure air source device of 14-, 15-capsule fixed block, 16-chain, the 17-LED light source, 18-soft capsule, 19-light box
The specific embodiment
As shown in Figure 1, checkout gear of the present invention is arranged transport portion, Computer Vision Recognition part and capsule by capsule and is rejected action part and form.The camera 5 of lining by line scan of Computer Vision Recognition part is installed in carrier chain 3 tops that capsule is arranged transport portion, and the air-blast atomizer 6 that capsule is rejected the action part installation is installed in carrier chain 5 sides that capsule is arranged transport portion.
The function that described capsule is arranged the transport portion realization is to make the online orderly arrangement of soft capsule, main involving vibrations feed appliance 2 and carrier chain 3.Vibration feeder 2 is installed in the front end of carrier chain 3, make 18 one-tenth single-row order outputs of software capsule, and be connected with carrier chain 3, in soft capsule 18-one the pit that drops on carrier chain 3 by the conduit of portion mounted thereto, this pit is formed by capsule fixed block 15, as shown in Figure 2.Soft capsule 18 is fed forward with certain intervening sequences level with carrier chain 3, in one one the visual field that is presented on the camera 5 of lining by line scan.
The function that described capsule is rejected the action part realization is to reject defective, the special-shaped soft capsules of those weight, mainly is linked in sequence by clean high-pressure air source 14, high speed miniature electromagnetic valve 13 and air-blast atomizer 6 and forms.Air-blast atomizer 6 is installed in the side of carrier chain 3 rear ends, requires them to produce action according to the graded signal that computer vision partly provides, and the soft capsule 18 that weight is defective, special-shaped is rejected from carrier chain 3 accurately.
Described Computer Vision Recognition part, it is core of the present invention, is triggered sensor 4, control module and corresponding software and is formed by a camera system of lining by line scan, lighting box 19, a photoelectricity.The system of taking the photograph of wherein lining by line scan is linked in sequence and is formed by a camera 5 of lining by line scan, a high speed image acquisition board, a computer 8.Photoelectricity triggers sensor 4 and is installed on the carrier chain 3, links to each other with control module with high speed image acquisition board in the computer 8 by circuit, and the triggering signal of image taking and the position of each capsule are provided.As shown in Figure 3, lighting box 19 is installed in the top of carrier chain 3, and the inside is mounted with the led light source 17 of bar shaped, produces on soft capsule 18 surfaces to diffuse uniformly.The camera 5 of lining by line scan is installed in the lighting box 19, and perpendicular to be installed in carrier chain 3 directly over, can photograph the image of soft capsule 18, the high-speed figure image pick-up card in camera 5 via line and the computer 8 links to each other, and deposits captured image in computer.Computer 8 links to each other with control module by data wire, send instruction to control module according to the result who handles, control module is placed in the switch board 9, control module is sent magnetic valve 13 actions that the signal controlling capsule is rejected action part, and the soft capsule 18 that weight is defective, special-shaped is rejected from carrier chain 3 accurately.Qualified soft capsule 18 enters in the pulp-collecting box automatically.The weight judgement of described soft capsule is to realize detecting by the characteristics of image parameter of setting up capsule and the relation (as shown in Figure 4) between the capsules weight.
During work, soft capsule 18 is arranged at capsule and is arranged in order on the transport portion and along with carrier chain 3 advances, partly touch photoelectricity in Computer Vision Recognition and trigger sensor 4, photoelectricity triggers sensor 4 the capsule position is sent to control module and computer 8, and the high speed image acquisition board in the triggering computer 8, the image of collection soft capsule 18.Computer 8 is in time handled captured sequence image after having taken at every turn, extracts characteristic parameters such as area, size, shape, finishes by identification software whether soft capsule 18 is the detection of special-shaped ball and weight.To reject instruction by control module at last and send to capsule rejecting action part, produce action, special-shaped ball and the underproof soft capsule 18 of weight will be rejected from carrier chain 3 accurately by magnetic valve 13.Qualified soft capsule 18 enters in the pulp-collecting box 7 automatically.