CN105825182A - Double analysis method of online glume-gaping seed rice identification - Google Patents
Double analysis method of online glume-gaping seed rice identification Download PDFInfo
- Publication number
- CN105825182A CN105825182A CN201610146354.6A CN201610146354A CN105825182A CN 105825182 A CN105825182 A CN 105825182A CN 201610146354 A CN201610146354 A CN 201610146354A CN 105825182 A CN105825182 A CN 105825182A
- Authority
- CN
- China
- Prior art keywords
- seed rice
- grain husk
- image
- splits
- analysis method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Abstract
The invention discloses a double analysis method of online glume-gaping seed rice identification. A monochrome industrial camera collects front-side and back-side images of moving seed rice synchronously, an image processing algorithm is used to carry out image processing and glume-gaping feature extraction on the front-side and back-side images successively, detection identification is carried out via the features in the front-side and back-side images, analysis results of the front-side and back-side images are integrated to implement screening and rejecting, and a detection result of the glume-gaping seed rice is obtained. According to the invention, the images of the two sides of the seed rice are obtained synchronously, multiple threads are used to process the images and realize engineering, and the glume-gaping seed rice can be identified and processed in higher precision and efficiency.
Description
Technical field
The present invention relates to technical field of machine vision, particularly relate to a kind of ONLINE RECOGNITION and split the two-sided analysis method of grain husk seed rice.
Background technology
Hybrid rice seed rice is commonly present and interior coetonium not completely closed splits grain husk defect, cause the vigor of seed rice and germination percentage to reduce, and existing selecting crude drugs with winnower process means cannot automatic sorting.Utilize the research of machine vision technique detection seed quality, including patent CN200710067204.7, CN201110122572.3, CN200410078033.4, CN200910148225.0, CN201010237646.3 etc., all use one side detection or asynchronous collection dual-side image, and seed rice attitude cannot be fixed, splitting grain husk and may be presented on not coplanar, prior art cannot realize the dual-side image of the same seed rice of synchronous acquisition and carry out splitting the grain husk online two-sided analysis of seed rice.
Summary of the invention
For solving the problems referred to above that prior art exists, the invention provides a kind of ONLINE RECOGNITION and split the two-sided analysis method of grain husk seed rice.
The technical solution used in the present invention is:
1) the front and back image of black and white industrial camera synchronous acquisition motion seed rice is utilized；
2) image processing algorithm is utilized direct picture and back side image to carry out image procossing respectively successively and splits grain husk feature extraction；
3) by the feature in direct picture and back side image carries out detection identification, the analysis result of last comprehensive dual-side image carries out screening and rejects, it is thus achieved that split the testing result of clever seed rice.
Described step 3) specifically: direct picture and back side image are carried out Hough straight-line detection respectively, and the quantitative determination of joint line splits grain husk seed rice or normal seed rice: if the quantity of line is less than or equal to 1, be then normal seed rice；If the quantity of line is equal to 2, the most again by the difference of the y-coordinate at two straight line midpoints in deviation threshold, then for being normal seed rice；If the quantity of line is more than or equal to 3, then for splitting grain husk seed rice；After dual-side image all detects, direct picture is identical with the testing result of back side image, then retain its testing result；Direct picture is different with the testing result of back side image, then it is assumed that this seed rice is for splitting grain husk seed rice
Described image procossing is to include removing image background, then by rotating and translating, motion seed rice is placed in picture centre.
Described image procossing is specifically: use Da-Jin algorithm that image is carried out Threshold segmentation, find maximum area connected domain again, it is seed rice region, maximum area connected domain is filled by white, other regions are set to 0 as background area and by its pixel value, finally utilize medium filtering (3*3) to remove picture noise point；Utilize cvMinAreaRect2 function to obtain white and fill the minimum enclosed rectangle in region, according to functional relationship between limit long in minimum enclosed rectangle and the angle α of horizontal axis and anglec of rotation β, seed rice major axis is rotated to horizontal direction centered by its boundary rectangle center, remainder pixel value is 0 to be filled with, then use edge detection operator to carry out detection and obtain contour edge feature, and remove.
Described grain husk feature extraction of splitting specifically includes the edge feature extracting motion seed rice, removes the profile of motion seed rice, it is thus achieved that the middle edge feature of motion seed rice.
Described step 1) use double camera correlation mode to obtain the two-sided picture rich in detail of same seed rice in motion simultaneously.
Described step 2) process and step 3) in detection identification process use multithreading to carry out image procossing respectively.
The image coordinate system of the present invention is with image straight down for y-axis positive direction, with image level to the right for x-axis positive direction.
Compared with existing seed quality detection technique, the device have the advantages that into:
Synchronizing to obtain the tow sides image of seed rice, effectively reduce the blind area of seed rice surface information and the object error of asynchronous collecting, accuracy of identification is high；Multithreading image procossing and the software of utilizing works code development, it is achieved ONLINE RECOGNITION splits grain husk seed rice in high volume, efficiency is high.
Accompanying drawing explanation
Fig. 1 is a kind of embodiment flow chart of the inventive method.
Fig. 2 is normal seed rice image.
Fig. 3 is to split grain husk seed rice image.
Fig. 4 is normal seed rice image after removal frame.
Fig. 5 is to remove to split grain husk seed rice image after frame.
Fig. 6 is normal seed rice binary image.
Fig. 7 is to split grain husk seed rice binary image.
Fig. 8 is normal seed rice image after removal background.
Fig. 9 is to remove to split grain husk seed rice image after background.
Figure 10 is normal seed rice boundary rectangle vertex graph.
Figure 11 is to split grain husk seed rice boundary rectangle vertex graph.
Figure 12 is normal seed rice binary map after rotation.
Figure 13 be rotate after split grain husk seed rice binary map.
Figure 14 is normal seed rice image after rotation.
Figure 15 be rotate after split grain husk seed rice image.
Figure 16 is normal seed rice rim detection binary map.
Figure 17 is to split grain husk seed rice rim detection binary map.
Figure 18 is to remove normal seed rice edge binary map after profile.
Figure 19 is to remove to split grain husk seed rice edge binary map after profile.
Figure 20 is to be detected as normal seed rice image.
Figure 21 is to be detected as splitting grain husk seed rice image.
Figure 22 be line segment quantity be the normal seed rice image of 1.
Figure 23 be line segment quantity be the normal seed rice image of 2.
Detailed description of the invention
Below in conjunction with the accompanying drawings and the present invention is described in further detail by specific embodiment.
Embodiments of the invention are as follows, comprise the following steps as shown in Figure 1:
S1, utilize the front and back image of black and white industrial camera synchronous acquisition motion seed rice, and be saved to the data buffer area of correspondence respectively.Data buffer area A and data buffer area B is size and the camera colour bits that match are the frame image data capacity of 8 deeply.
S2, monitored data buffer area, check whether data buffer area A or B is filled, and arbitrary buffer area is filled then two threads of unlatching.Data buffer area A is filled one thread process seed rice direct picture of use, and data buffer area B is filled another thread process seed rice back side image of use.
S3, in respective thread, carry out image procossing and split grain husk feature extraction, carrying out the quantitative determination of Hough straight-line detection joint line afterwards and split grain husk seed rice or normal seed rice, respectively obtain respective analysis result.Concrete processing procedure is as follows:
The normal seed rice image that collects as in figure 2 it is shown, split grain husk seed rice as it is shown on figure 3, first remove the impact of image upper and lower side frame, will image longitudinal coordinate direction 0～116,605～640 row pixel value a little be all set to 0, respectively obtain Fig. 4 and Fig. 5.
Use Da-Jin algorithm that Fig. 4 and Fig. 5 carries out Threshold segmentation, then find maximum area connected domain, be seed rice region, fill maximum area connected domain by white, finally utilize medium filtering (3*3) to remove picture noise point, obtain Fig. 6 and Fig. 7；
Retain target area in Fig. 4 and Fig. 5, the white portion part corresponding to Fig. 6 and Fig. 7 respectively, and its background area pixels value is set to 0, respectively obtain Fig. 8 and Fig. 9.
Utilizing cvMinAreaRect2 function to fill the minimum enclosed rectangle in region to obtain Fig. 6 and Fig. 7 white, return value is CvBox2D structure, and this structure includes boundary rectangle centre coordinate, rectangle length and wide, rectangle Article 1 limit and the angle α of horizontal axis.Seed rice major axis is rotated to horizontal direction centered by its boundary rectangle center, the functional relationship of α and anglec of rotation β need to be known.
Available cvBoxPoints function obtains the coordinate on four summits of rectangle, finding first point is that (initial point is in the upper left corner for coordinate y, being y straight down, level is to the right for x) being that (Figure 10 and Figure 11 midpoint 1) that in four summits, y value is maximum.Then the point found, according to clockwise, is followed successively by 2, and 3,4 represent remaining rectangle summit.D1 represents the distance between first summit and second summit, and d2 represents the distance between second summit and the 3rd summit, compares d1, d2 size and determines functional relationship between α and β:
β is made the input parameter of cv2DrotationMatrix function.Result is as shown in Figure 12 and Figure 13.
By Fig. 8 and Fig. 9 seed rice image, according to they minimum enclosed rectangle, rotating around minimum enclosed rectangle center, rotate the major axis of seed rice to horizontal level, remainder pixel value is 0 to be filled with；Then by the center of minimum enclosed rectangle center translation to whole image, Figure 14 and 15 is obtained.
The gray value splitting grain husk subregion image after splitting grain husk seed rice surface imaging is of a relatively high, and it and surrounding normal parts of images show the sudden change of gray scale, therefore edge detection operator can be used to carry out feature extraction.Common edge detective operators includes Sobel, Prewitt, Log, Canny, contrast this several operators, Sobel and Prewitt calculates simple, the general profile of image can only be detected, and to crackle defects detection poor effect, Log and Canny is capable of detecting when thin edge, but Canny has more preferable Detection results and is susceptible to the interference of noise.Using Canny operator that Figure 14 and Figure 15 is carried out rim detection, result is as shown in FIG. 16 and 17.
Knowable to Figure 16 and Figure 17, splitting grain husk part line-like, but seed rice profile has interference to follow-up straight-line detection, with Figure 16 and Figure 17 dot product respectively, Figure 12 and Figure 13 can be removed seed rice profile after corrosion, result is as shown in Figure 18 and Figure 19；
Knowable to Figure 18 and 19, split grain husk near linear, it is also contemplated that the efficiency of on-line checking.Detect with the conversion of statistical probability Hough line and split grain husk straight line (testing result is as shown in Figure 20 and Figure 21), during defer to certain criterion: one is, split grain husk and sterile lemma line segment is all straight line, the former level, the latter has certain angle of inclination, line segment angle of inclination can be controlled, angle absolute value threshold value 10 degree corresponding to line segment slope is set herein；Two are, with the presence of shorter line segment (as shown in figure 18), it is clear that this position is not to split grain husk position, therefore short segment is rejected, and threshold value is set to 15；Three are, in operating process, normal seed rice is also possible to there is line segment, determine normal by the quantity of alternative line or split grain husk, it is normal seed rice (Figure 22) when line segment quantity is less than or equal to 1, when line segment quantity is equal to 2, may be normal, may be for splitting grain husk, pass through laboratory observation, if splitting grain husk, then the y-coordinate at these two straight line midpoints should be sufficiently close to, and it is normal (Figure 23) apart from each other, distinguish normal by setting deviation threshold or split grain husk, when straight line quantity is more than or equal to 3, for splitting grain husk seed rice (Figure 21).
S4, comprehensive two-sided analysis result, i.e. the two-sided information of seed rice, finally judges.Result of determination is sent to grading plant, and grading plant is rejected according to result of determination and is split grain husk seed rice.Meanwhile, return monitored data buffer area, carry out the identification of next seed rice.
The final experimental result of the present embodiment is as shown in the table, it is seen that it has prominent significant technique effect, effectively reduces the blind area of seed rice surface information and the object error of asynchronous collecting, and accuracy of identification is high, and efficiency is high:
| Quantity | Correctly identify quantity | Correct recognition rata | |
| Normal seed rice | 453 | 423 | 93.1% |
| Split grain husk seed rice | 554 | 501 | 90.4% |
The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. made, should be included within the scope of the present invention.
Claims (7)
1. an ONLINE RECOGNITION splits the two-sided analysis method of grain husk seed rice, it is characterised in that comprise the following steps:
1) the front and back image of black and white industrial camera synchronous acquisition motion seed rice is utilized；
2) image processing algorithm is utilized direct picture and back side image to carry out image procossing respectively successively and splits grain husk feature extraction；
3) by the feature in direct picture and back side image carries out detection identification, the analysis result of last comprehensive dual-side image carries out screening and rejects, it is thus achieved that split the testing result of clever seed rice.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterized in that: described step 3) specifically: direct picture and back side image are carried out Hough straight-line detection respectively, the quantitative determination of joint line splits grain husk seed rice or normal seed rice: if the quantity of line is less than or equal to 1, be then normal seed rice；If the quantity of line is equal to 2, the most again by the difference of the y-coordinate at two straight line midpoints in deviation threshold, then for being normal seed rice；If the quantity of line is more than or equal to 3, then for splitting grain husk seed rice；
After dual-side image all detects, direct picture is identical with the testing result of back side image, then retain its testing result；Direct picture is different with the testing result of back side image, then it is assumed that this seed rice is for splitting grain husk seed rice.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterised in that: described image procossing is to include removing image background, then by rotating and translating, motion seed rice is placed in picture centre.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterized in that: described image procossing specifically: use Da-Jin algorithm image is carried out Threshold segmentation, find maximum area connected domain again as seed rice region, maximum area connected domain is filled by white, other regions are set to 0 as background area and by its pixel value, finally utilize medium filtering (3*3) to remove picture noise point；Utilize cvMinAreaRect2 function to obtain white and fill the minimum enclosed rectangle in region, according to functional relationship between limit long in minimum enclosed rectangle and the angle α of horizontal axis and anglec of rotation β, seed rice major axis is rotated to horizontal direction centered by its boundary rectangle center, remainder pixel value is 0 to be filled with, then use edge detection operator to carry out detection and obtain contour edge feature, and remove.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterised in that: described grain husk feature extraction of splitting specifically includes the edge feature extracting motion seed rice, removes the profile of motion seed rice, it is thus achieved that the middle edge feature of motion seed rice.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterised in that: described step 1) use double camera correlation mode to obtain the two-sided picture rich in detail of same seed rice in motion simultaneously.
ONLINE RECOGNITION the most according to claim 1 splits the two-sided analysis method of grain husk seed rice, it is characterised in that: described step 2) process and step 3) and in detection identification process use multithreading to carry out image procossing respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610146354.6A CN105825182A (en) | 2016-03-15 | 2016-03-15 | Double analysis method of online glume-gaping seed rice identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610146354.6A CN105825182A (en) | 2016-03-15 | 2016-03-15 | Double analysis method of online glume-gaping seed rice identification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105825182A true CN105825182A (en) | 2016-08-03 |
Family
ID=56987939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610146354.6A Pending CN105825182A (en) | 2016-03-15 | 2016-03-15 | Double analysis method of online glume-gaping seed rice identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105825182A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109632007A (en) * | 2019-01-17 | 2019-04-16 | 北京理工大学 | A kind of edge point extracting method and gear high-precision vision measuring system |
| CN109961441A (en) * | 2019-03-19 | 2019-07-02 | 广东省农业科学院水稻研究所 | Hybrid rice seed splits the efficient measuring method of clever rate |
| CN110501065A (en) * | 2019-07-24 | 2019-11-26 | 南京农业大学 | Hybrid rice based on collision characteristic splits clever Seed inspection method |
| CN113305017A (en) * | 2021-05-28 | 2021-08-27 | 柳州源创电喷技术有限公司 | Comprehensive intelligent detection and sorting method for full-automatic valve element |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1428602A (en) * | 2001-08-23 | 2003-07-09 | 株式会社安西总合研究所 | Grain discriminating device and method |
| US20050226465A1 (en) * | 2004-04-09 | 2005-10-13 | Nichizo Iron Works & Marine Corporation. Shoei Foods Corporation, And Aio. , Ltd. | Seed fragment inspection apparatus |
| CN101929961A (en) * | 2009-06-18 | 2010-12-29 | 华东交通大学 | Device and method for detecting quality of rice seeds, identifying varieties and grading |
-
2016
- 2016-03-15 CN CN201610146354.6A patent/CN105825182A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1428602A (en) * | 2001-08-23 | 2003-07-09 | 株式会社安西总合研究所 | Grain discriminating device and method |
| US20050226465A1 (en) * | 2004-04-09 | 2005-10-13 | Nichizo Iron Works & Marine Corporation. Shoei Foods Corporation, And Aio. , Ltd. | Seed fragment inspection apparatus |
| CN101929961A (en) * | 2009-06-18 | 2010-12-29 | 华东交通大学 | Device and method for detecting quality of rice seeds, identifying varieties and grading |
Non-Patent Citations (3)
| Title |
|---|
| PEARSON T.ET AL: "《A machine vision system for high speed sorting of small spots on grains》", 《JOURNAL O FOOD MEASUREMENT & CHARACTERIZATION》 * |
| 成芳: "《稻种质量的机器视觉无损检测研究》", 《中国优秀博硕士学位论文全文数据库（博士） 信息科技辑》 * |
| 成芳: "《稻种质量的机器视觉无损检测研究》", 《中国优秀博硕士学位论文全文数据库（博士） 农业科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109632007A (en) * | 2019-01-17 | 2019-04-16 | 北京理工大学 | A kind of edge point extracting method and gear high-precision vision measuring system |
| CN109632007B (en) * | 2019-01-17 | 2020-12-04 | 北京理工大学 | Edge point extraction method and gear high-precision vision measurement system |
| CN109961441A (en) * | 2019-03-19 | 2019-07-02 | 广东省农业科学院水稻研究所 | Hybrid rice seed splits the efficient measuring method of clever rate |
| CN110501065A (en) * | 2019-07-24 | 2019-11-26 | 南京农业大学 | Hybrid rice based on collision characteristic splits clever Seed inspection method |
| CN113305017A (en) * | 2021-05-28 | 2021-08-27 | 柳州源创电喷技术有限公司 | Comprehensive intelligent detection and sorting method for full-automatic valve element |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111292305B (en) | Improved YOLO-V3 metal processing surface defect detection method | |
| CN101995223B (en) | Chip appearance detection method and system | |
| CN107966454A (en) | A kind of end plug defect detecting device and detection method based on FPGA | |
| CN112862770B (en) | Defect analysis and diagnosis system, method and device based on artificial intelligence | |
| CN105825182A (en) | Double analysis method of online glume-gaping seed rice identification | |
| CN111753692A (en) | Target object extraction method, product detection method, device, computer and medium | |
| CN106290392A (en) | A kind of little micro-bearing surface pitting defects online test method and system thereof | |
| US6993187B2 (en) | Method and system for object recognition using fractal maps | |
| CN110648330B (en) | Defect detection method for camera glass | |
| CN106780437B (en) | A kind of quick QFN chip plastic packaging image obtains and amplification method | |
| CN112837308A (en) | Building crack detection method, device, equipment and storage medium | |
| US11068740B2 (en) | Particle boundary identification | |
| CN102901735A (en) | System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer | |
| CN110060239B (en) | Defect detection method for bottle opening of bottle | |
| CN115170518A (en) | Cell detection method and system based on deep learning and machine vision | |
| CN113916893A (en) | Method for detecting die-cutting product defects | |
| CN112345534B (en) | Defect detection method and system for particles in bubble plate based on vision | |
| CN117252861A (en) | Method, device and system for detecting wafer surface defects | |
| CN116433978A (en) | Automatic generation and automatic labeling method and device for high-quality flaw image | |
| CN113870754B (en) | Method and system for judging defects of panel detection electronic signals | |
| CN110619273B (en) | Efficient iris recognition method and recognition device | |
| CN103593667A (en) | Rapid image foreign matter identification method based on set connectivity principle | |
| CN109636778B (en) | Defect detection method and defect detection device for display panel | |
| KR101646838B1 (en) | Counting method for track images | |
| CN117495846B (en) | Image detection method, device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160803 |