Embodiment
By describing technology contents of the present utility model in detail, realized object and effect, accompanying drawing is coordinated to be explained below in conjunction with embodiment.
The design of the utility model most critical is: if by glass out-of-flatness, to the principle of distortion be there is in the image then projected on it, use image recognition processing technology, automatic identification also orients the place of pattern deformation, and what realize high measurement accuracy and low cost realizes the high efficiency flatness defect detecting glass.
The flexibility flatness of parallel goods and the goodness of fit (GB)
Refer to Fig. 3 and Fig. 9, the utility model provides a kind of glass flatness pick-up unit, comprises lamp source 1, light-passing board 2 and glass to be detected 3, and described light-passing board 2 is provided with the pattern be made up of the pattern arrangement of multiple same specification;
Described lamp source 1, light-passing board 2, and the direction that glass to be detected 3 penetrates along lamp source 1 is arranged in order setting;
Also comprise interconnective photographic means 4 and image processor 5; Described image processor 5 comprises interconnective testing circuit 51 and positioning circuit 52; Described testing circuit 51 is connected with described photographic means 4;
Described photographic means 4, is incident upon for obtaining described light-passing board 2 image that described glass to be detected 3 is formed;
Described testing circuit 51, for identifying the described figure that described image is out of shape;
Described positioning circuit 52, for the position of the figure of positioning transformations.
Further, described pattern is lattice; Above-mentioned photographic means 4 is preferably CCD photographing device, and described light source is preferably LED light source.
From foregoing description, the beneficial effects of the utility model are: above-mentioned detection device is mainly used in glass production streamline and detects, and can reduce manpower, get rid of naked eyes simultaneously and glass planarization cannot be detected and the optical distortion that causes, affect the quality of glass.When intelligent light source is radiated on glass 3 to be detected through light-passing board 2, lattice on light-passing board 2 can reflex to the wall on opposite, the image being incident upon and glass 3 to be detected generates can be obtained by CCD photographing device, if glass surface is smooth, all grid patterns then on lattice are all indeformable, otherwise existence is out of shape by grid image; Concrete detection mode can by image pick-up card image data acquiring in calculator memory, by image detection device, Aulomatizeted Detect analytic process is carried out to collected image again, realize the detection to glass surface flatness defect, finally by output device output detections result.
Further, described testing circuit 51 comprises the sign unit, arithmetic element, comparing unit and the acquiring unit that connect successively;
Described unit, for indicating each grid pattern in described grid image;
Described arithmetic element, for calculating the area of each grid pattern;
Described comparing unit, for the area of single grid pattern on the area of more each grid pattern and described light-passing board 2;
Described acquiring unit, for obtaining the unequal grid pattern of area with single grid on described light-passing board 2 on described grid image.
Seen from the above description, by the above-mentioned Obtaining Accurate being testing circuit 51 and achieving to the figure obtained image existing deformation, the precision and efficiency of detecting of glass defect is greatly improved.
Further, described image processor 5 also comprises interconnective pre-process circuit 53 and binary conversion treatment circuit 54; Described pre-process circuit 53 is also connected with photographic means 4, and described binary conversion treatment circuit 54 is also connected with testing circuit 51;
Described pre-process circuit 53, for carrying out pre-service to described image;
Described binary conversion treatment circuit 54, for carrying out binary conversion treatment to pretreated image.
Further, described image processor 5 also comprises boundary treatment circuit 55 and convergent-divergent circuit 56, and described binary conversion treatment circuit 54, boundary treatment circuit 55, convergent-divergent circuit 56 are connected successively with testing circuit 51;
Described boundary treatment circuit 55, for carrying out boundary treatment to the image after described binary conversion treatment, removes the part beyond described pattern coverage on described image;
Described convergent-divergent circuit 56, for scaled described image.
Seen from the above description, by above-mentioned pre-process circuit 53 and binary conversion treatment circuit 54, achieve the remarkable enhancing of the detectability to accessed image, improve accuracy of detection and the detection efficiency of the figure of distortion.
Please refer to Fig. 1, Fig. 2 and Fig. 3, based on a kind of method that above-mentioned detection device provides glass flatness to detect, comprise lamp source 1, light-passing board 2 and glass to be detected 3, described light-passing board 2 is provided with the pattern be made up of the pattern arrangement of multiple same specification;
Described lamp source 1, light-passing board 2, and the direction that glass to be detected 3 penetrates along lamp source 1 is arranged in order setting;
Obtain described light-passing board 2 and be incident upon the image that described glass to be detected 3 is formed;
Identify the described figure that described image is out of shape;
The position of the figure of positioning transformations.
Further, pattern is lattice.Certainly, described pattern can also be made up of the triangle of other same size specifications, circle, polygonal arrangement, only need guarantee that the figure forming pattern is unified shape, size, and be evenly distributed.
From the above, the beneficial effect of above-mentioned detection method is: above-mentioned detection method is mainly used in glass production streamline and detects, and can reduce manpower, gets rid of naked eyes simultaneously and glass planarization cannot be detected and the optical distortion that causes, affect the quality of glass.When light source is radiated on glass 3 to be detected through light-passing board 2, lattice on light-passing board 2 can reflex to the wall on opposite, the image being incident upon and glass 3 to be detected generates can be obtained by camera head, if glass surface is smooth, all grid patterns then on lattice are all indeformable, otherwise existence is out of shape by grid image; Concrete detection mode can by image pick-up card image data acquiring in calculator memory, by image detection device, Aulomatizeted Detect analytic process is carried out to collected image again, realize the detection to glass surface flatness defect, finally by output device output detections result.
Above-mentioned detection method effectively can solve and detect glass flatness defect by artificial means, even for tiny distortion, reaches problem that is quick, efficient detection glass flatness defect.
Further, described " identifying the described figure that described image is out of shape " is specially:
Indicate each grid pattern in described grid image, calculate the area of each grid pattern;
The area of single grid pattern on the area of more each grid pattern and described light-passing board 2;
Obtain the unequal grid pattern of area with single grid on described light-passing board 2 on described grid image.
Seen from the above description, robotization can be realized by said process and identify in image the grid pattern that there is distortion, improve detection efficiency.
Further, after described " obtain described light-passing board 2 and be incident upon the image that described glass to be detected 3 is formed ", comprise further:
Pre-service is carried out to described image; Binary conversion treatment is carried out to pretreated image.
Refer to Fig. 6 and Fig. 7, further, describedly " pre-service is carried out to described image; Binary conversion treatment is carried out to pretreated image " be specially:
Histogram equalization, geometric transformation or image smoothing method is used to carry out pre-service to described image;
Binary conversion treatment is carried out to described pretreated image;
Boundary treatment is carried out to the image after described binary conversion treatment, removes the part beyond described pattern coverage on described image;
Scaled described image.
Seen from the above description, first can carry out Image semantic classification by the above-mentioned image to directly getting, the noise in removal of images transmitting procedure, realizing removing the irrelevant information in image; By truly useful Information recovering, the contrast of pattern can be strengthened again simultaneously, strengthen the detectability of image further; And the binary conversion treatment process of image, image is converted to monochrome presentation, make image procossing become simple, reduce data volume simultaneously, image goes out objective contour; Further, further comprises the processing procedure to image boundary, non-targeted object is removed, and the scale smaller of image, realize the reduction of data to be tested, concentrating of target detection object; By said process, finally realize the raising of accuracy of detection and detection efficiency.
Please refer to Fig. 1 to Fig. 8, embodiment one of the present utility model is:
The method that glass flatness detects, adopts advanced CCD imaging technique and intelligent light source, specifically comprises LED, the light-passing board 2 of good light permeability, CCD camera and the image processor 5 with good transmission performance; Described light-passing board 2 is provided with by multiple pattern with unified specification figure proper alignment, or be directly the hollowed-out translucent plate 2 of above-mentioned pattern, described figure can be triangle, circle, rectangle or polygon, at this, with described pattern for lattice is described.
Described lamp source 1, light-passing board 2, and the direction that glass to be detected 3 penetrates along lamp source 1 is arranged in order setting, to be detected of described glass 3 to be detected is relative with lamp source 1; Being provided with smooth projection wall 6 after preferred glass to be measured, can be directly wall;
Lamp source 1 is irradiated on glass 3 to be detected through light-passing board 2, lattice on light-passing board 2 is radiated at rear through glass 3 to be detected and lays on the projection wall 6 of glass 3 to be detected, by reflection, formed grid image is projected on the reception reflecting wall 7 on opposite; The grid image received on reflecting wall 7 is obtained by CCD camera, as shown in Figure 5, imported in computing machine, after the acquisition grid image of computing machine, processed described grid image by the image processor 5 of its configuration, concrete processing procedure can comprise:
1, Image semantic classification and binary conversion treatment process;
Histogram equalization, geometric transformation or image smoothing method is used to carry out pre-service to described image; By the noise in preprocessing process removal of images transmitting procedure, realize removing the irrelevant information in image; By truly useful Information recovering, can strengthen the contrast of pattern again, strengthen the detectability of image further, the effect after process as shown in Figure 6 simultaneously;
Binary conversion treatment is carried out to described pretreated image; Convert image to monochrome presentation by binary conversion treatment, make image procossing become simple, reduce data volume simultaneously, image goes out objective contour and shows;
Boundary treatment is carried out to the image after described binary conversion treatment, removes the part beyond described pattern coverage on described image; By boundary treatment, the non-targeted object on image is removed, as shown in Figure 6;
Scaled described image; By boundary treatment and reduce process, realize the reduction of data to be tested, concentrating of target detection object.
2, the described figure that described image is out of shape is identified;
Each grid pattern in two-dimentional binary image after above-mentioned process is indicated; Calculate the area of each grid pattern marked;
With the area of the grid pattern in the normal lattice on light-passing board 2 for foundation, compare the size of the area of normal single grid pattern on the area of each grid pattern marked and described light-passing board 2 successively;
Obtain the unequal grid pattern of area with single grid on described light-passing board 2 on described grid image, namely identify the grid pattern of distortion, as shown in Figure 7.
At this, also figure can be adopted to cover the mode compared and to identify choppy, specifically comprise the grid pattern obtained in normal lattice, as standard grid pattern, successively each grid pattern in described two-dimentional binary image and standard grid pattern are carried out covering comparison, determine the grid pattern of distortion with this.
3, the position of the figure of positioning transformations.
According to the figure of the position positioning transformations indicated, and store the graph position of distortion, to reach the object of location, as shown in Figure 8, at the second row secondary series, glass surface out-of-flatness produces distortion.
Please refer to Fig. 3 to Fig. 9, embodiment two of the present utility model is:
A kind of glass flatness pick-up unit, comprise LED lamp source 1, light-passing board 2 and glass to be detected 3, described light-passing board 2 is provided with the pattern be made up of the pattern arrangement of multiple same specification, and preferred described pattern is lattice;
Described lamp source 1, light-passing board 2, and the direction that glass to be detected 3 penetrates along lamp source 1 is arranged in order setting;
Also comprise interconnective CCD photographic means 4 and image processor 5; Described image processor 5 comprises interconnective pre-process circuit 53, binary conversion treatment circuit 54, boundary treatment circuit 55, convergent-divergent circuit 56, testing circuit 51 and positioning circuit 52; Described pre-process circuit 53 is connected with described CCD photographic means 4; Described testing circuit 51 comprises the sign unit, arithmetic element, comparing unit and the acquiring unit that connect successively; Described acquiring unit is connected with described positioning circuit 52.
Described photographic means 4, is incident upon for obtaining described light-passing board 2 image that described glass to be detected 3 is formed;
Described testing circuit 51, for identifying the described figure that described image is out of shape;
Described positioning circuit 52, for the position of the figure of positioning transformations;
Described pre-process circuit 53, for carrying out pre-service to described image; Histogram equalization, geometric transformation or image smoothing method specifically can be used to carry out pre-service to described image;
Described binary conversion treatment circuit 54, for carrying out binary conversion treatment to pretreated image;
Described boundary treatment circuit 55, for carrying out boundary treatment to the image after described binary conversion treatment, removes the part beyond described pattern coverage on described image;
Described convergent-divergent circuit 56, for scaled described image;
Described sign unit, for indicating each grid pattern in described grid image;
Described arithmetic element, for calculating the area of each grid pattern;
Described comparing unit, for the area of single grid pattern on the area of more each grid pattern and described light-passing board 2;
Described acquiring unit, for obtaining the unequal grid pattern of area with single grid on described light-passing board 2 on described grid image.
In sum, the device that a kind of glass flatness that the utility model provides detects, if by glass out-of-flatness, to the principle of distortion be there is in the image then projected on it, use image recognition processing technology, automatic identification also orients the place of pattern deformation, and what realize high measurement accuracy and low cost realizes the high efficiency flatness defect detecting glass; Further, by carrying out pre-service, binary conversion treatment, boundary treatment and scale smaller process to accessed image, realize the reduction of view data to be detected, concentrating of target detection object, finally realizes the accuracy of detection of glass flatness defect to be detected and the raising of detection efficiency.
The foregoing is only embodiment of the present utility model; not thereby the scope of the claims of the present utility model is limited; every equivalents utilizing the utility model instructions and accompanying drawing content to do; or be directly or indirectly used in relevant technical field, be all in like manner included in scope of patent protection of the present utility model.