CN105447859A - Field wheat aphid counting method - Google Patents
Field wheat aphid counting method Download PDFInfo
- Publication number
- CN105447859A CN105447859A CN201510792807.8A CN201510792807A CN105447859A CN 105447859 A CN105447859 A CN 105447859A CN 201510792807 A CN201510792807 A CN 201510792807A CN 105447859 A CN105447859 A CN 105447859A
- Authority
- CN
- China
- Prior art keywords
- image
- aphid
- original image
- wheat
- counting
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Abstract
The invention discloses a field wheat aphid counting method. The method comprises the following steps of image acquisition: when wheat is grown into a booting stage from a jointing stage, acquiring a wheat-plant original image through a digital camera, acquiring X small pixel subimages in the original image and taking as a training sample; training sample collection: randomly intercepting N aphid images from the original image and taking as a positive sample, and randomly intercepting M images and taking as a negative sample; identification model training: extracting HOG characteristics of the positive sample and the negative sample, inputting the HOG of the positive sample and the negative sample into a SVM classifier to carry out training so as to obtain an insect pest classification model, using the samples to rectify the model and verifying an insect pest identification model; aphid detection: searching a MSER in the original image and finally using the SVM classifier to search aphids in a MSER area and record an aphid quantity. Operation is simple and aphid quantity calculating is high-efficient and accurate.
Description
Technical field
The present invention relates to a kind of method of counting, particularly a kind of field wheat aphid method of counting.
Background technology
Insect pest is the principal element affecting crop growth, and aphid is the normal insect sending out and retransmit in Wheat Production, not only draws plant juice, affect Wheat Development, can also propagate multiple virosis during its harm wheat.Accurate estimation pest density is the basis of carrying out prediction of pest attack forecast, and the timely acquisition of pest density carries out number of pest dynamically and the basis of pest damage degree analyzing, is equally also the necessary condition of acquisition Economic Threshold of Injurious Insect Control.At present more common aphid method of counting is artificial counting, because aphid individuality is less, observe that not only labor capacity is large, efficiency is low for a long time with eye, and counting accuracy rate is subject to the impact of subjective factor.Meanwhile, also there is a kind of automanual aphid method of counting, aphid is induced on specific plank by this method, utilizes graphical analysis to count.Usually, artificial counting method is divided into two kinds, and 1) sample and manually check the quantity of aphid on monolithic leaf, be then multiplied by every strain number of blade, then be multiplied by every square meter strain number; 2) aphid is induced on specific plank, by manually calculating Aphed population.Existing Intelligent counting method is induced to by aphid on specific plank, obtains plank image, then utilize image analysis technology, calculates aphid number.
Artificial counting is less due to aphid individuality, observes that not only labor capacity is large, efficiency is low for a long time with eye, and counting accuracy rate is subject to the impact of subjective factor; And specific plank counting operation trouble and be difficult to all aphids to be all induced on plank, there is comparatively big error.
Summary of the invention
Technical matters to be solved by this invention is to provide a kind of field wheat aphid method of counting, and it is simple to operate and can detect field wheat aphid quantity efficiently and accurately.
For solving the problems of the technologies described above, the technical solution adopted in the present invention is:
A kind of field wheat aphid method of counting, is characterized in that comprising following steps:
A, Image Acquisition: in wheat jointing to boot stage, obtain wheat plant original image by digital camera, obtain small pixel subimage X and open as training sample in original image;
B, training sample are collected: from original image, the random N that intercepts opens aphid image as positive sample, and the random M that intercepts opens image as negative sample;
C, model of cognition are trained: the HOG feature extracting positive sample and negative sample, the HOG of positive negative sample are input to SVM classifier and carry out training and obtain insect pest disaggregated model, correct with sample to model, checking insect model of cognition;
The detection of D, aphid: find MSER in original image, final SVM classifier is found aphid and is recorded Aphed population in MSER region;
Wherein, X, N, M are the natural number determined as required, and the value of M is greater than the value of N and the value of M, N is all less than X.
Further, described small pixel subimage size is 30*20 pixel.
Further, described wheat plant original image utilizes filtering operation to carry out pre-service to original image after obtaining, and eliminates impurity effect.
Further, described X value is that 5000-7000, N value is greater than 2000, M value and is greater than 5000.
Further, described find in original image MSER utilize formula q (i)=| Qi+ △-Qi-△ |/| Qi| carries out, a certain connected region when Qi represents that threshold value is i, △ is the change of gray threshold, q (i) is the rate of change of region Qi when threshold value is i, when q (i) is for local minimum, then Qi is maximum stable extremal region.
Further, described HOG characteristic extraction procedure is, by the original image gray processing of colour, utilizes formula I (x, y)=I (x, y)
gammacarry out image standardization, wherein gamma gets 0.5, the gradient of each pixel in computed image, image is divided into several 6 × 6 pixel cell lattice and the histogram of gradients of adding up each cell forms the descriptor of each cell, every 3 × 3 cells are formed a block, it is exactly the HOG descriptor of this block that the descriptor of the cell in a block is together in series, and in image, the HOG Feature Descriptor of all pieces is together in series is exactly the HOG Feature Descriptor of this image.
Further, the gradient calculation method of described each pixel is,
, G
x(x, y), G
y(x, y), H (x, y) represents the horizontal direction gradient at input picture (x, y) place, vertical gradient and pixel value respectively,
,
, G (x, y) is gradient magnitude, and α (x, y) is gradient direction.
Further, Haar feature detection is carried out after the positive and negative sample extraction of described HOG feature.
Further, described Haar feature detection comprises, and by edge feature, linear feature and central feature composition characteristic template, calculates Haar eigenwert by formula v=white pixel sum-2* black picture element sum.
The present invention compared with prior art, has the following advantages and effect: utilize computer picture method to measure the number of aphid on wheat plant, compare Traditional Man determination methods saving of work and time, and the investigation of Aphed population is not simultaneously by the impact of human factor; Compare the input that automanual image-recognizing method decreases labour equally, operation is more simple, improves efficiency and the aspect of investigation.
Accompanying drawing explanation
Fig. 1 is of the present invention for training the positive sample image of the part of SVM model.
Fig. 2 is of the present invention for training the part negative sample image of SVM model.
Fig. 3 is feature extraction design sketch of the present invention.
Embodiment
Below in conjunction with accompanying drawing, also by embodiment, the present invention is described in further detail, and following examples are explanation of the invention and the present invention is not limited to following examples.
A kind of field of the present invention wheat aphid method of counting, comprises following steps:
A, Image Acquisition: in wheat jointing to boot stage, obtain wheat plant original image by digital camera, image size is 1800*1200 pixel, obtains image and mainly comprises wheat plant itself and aphid, and complicated background can reduce the accuracy rate of identification.Filtering operation is utilized to carry out pre-service to original image, with the deimpurity impact that disappears.Then in original image, the small pixel subimage 6000 of 30*20 pixel is obtained as training sample by PS software.
B, training sample are collected: as shown in Figure 1, 2, and from original image, random intercepting is greater than the aphid image of 2000 as positive sample, and random intercepting is greater than 5000 images as negative sample, and positive and negative sample image is 30 × 20 pixels, is used for setting up SVM.
C, model of cognition are trained: the HOG feature extracting positive sample and negative sample, and carry out Haar feature detection, the HOG of positive negative sample are input to SVM classifier and carry out training and obtain insect pest disaggregated model, correct, checking insect model of cognition with sample to model.HOG characteristic extraction procedure comprises: by the original image gray processing of colour, utilizes formula I (x, y)=I (x, y)
gammacarry out image standardization, wherein gamma gets 0.5, utilize the gradient of each pixel in formula (1) ~ (3) computed image, image is divided into several 6 × 6 pixel cell lattice and the histogram of gradients of adding up each cell forms the descriptor of each cell, every 3 × 3 cells are formed a block, it is exactly the HOG descriptor of this block that the descriptor of the cell in a block is together in series, and in image, the HOG Feature Descriptor of all pieces is together in series is exactly the HOG Feature Descriptor of this image.
G
x(x,y)=H(x+1,y)-H(x-1,y)
G
y(x,y)=H(x,y+1)-H(x,y-1)(1)
Gx (x, y), Gy (x, y) in formula, H (x, y) represents the horizontal direction gradient at input picture (x, y) place, vertical gradient and pixel value respectively,
(2)
(3)
G (x, y) is gradient magnitude, and α (x, y) is gradient direction.
Haar feature is divided three classes: edge feature, linear feature, central feature, be combined into feature templates.Adularescent and black two kinds of rectangles in feature templates, and the eigenwert defining this template be white rectangle pixel and deduct black rectangle pixel and, Haar eigenwert reflects the grey scale change situation of image.Haar character numerical value computing formula is:
V=white pixel sum-black picture element sum; And for white in template being the situation of black twice, computing formula is as follows: v=white pixel number-2 × black picture element sum.
The detection of D, aphid: find MSER in original image, MSER is the most stable region obtained when using different gray thresholds to carry out binaryzation to image, be considered to the Affinely invariant region that performance is best, this algorithm may be used for the characteristic area under detection different visual angles and illumination condition.Utilize the MSER in formula (4) detected image,
q(i)=|Q
i+△-Q
i-△|/|Q
i|(4)
Wherein, a certain connected region when Qi represents that threshold value is i, △ is the change of gray threshold, and q (i) is the rate of change of region Qi when threshold value is i.When q (i) is for local minimum, then Qi is maximum stable extremal region.As shown in Figure 3, final SVM classifier is found aphid and is recorded Aphed population in MSER region.
The present invention utilizes computer picture method to measure the number of aphid on wheat plant, compares Traditional Man determination methods saving of work and time, and the investigation of Aphed population is not simultaneously by the impact of human factor.Compare the input that automanual image-recognizing method decreases labour equally, improve efficiency and the aspect of investigation.The present invention is at different densities, and different colours, difference depends on the aphid of position carries out Count Test, and average detected rate is more than 85%, and false detection rate is lower than 15%, and this has positive effect to the investigation that a situation arises of wheatland aphid.
Above content described in this instructions is only made for the present invention illustrating.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment; only otherwise depart from the content of instructions of the present invention or surmount this scope as defined in the claims, protection scope of the present invention all should be belonged to.
Claims (9)
1. a field wheat aphid method of counting, is characterized in that comprising following steps:
Image Acquisition: in wheat jointing to boot stage, obtains wheat plant original image by digital camera, obtains small pixel subimage X and open as training sample in original image;
Training sample is collected: from original image, the random N that intercepts opens aphid image as positive sample, and the random M that intercepts opens image as negative sample;
Model of cognition is trained: the HOG feature extracting positive sample and negative sample, the HOG of positive negative sample is input to SVM classifier and carries out training and obtain insect pest disaggregated model, correct with sample to model, checking insect model of cognition;
The detection of aphid: find MSER in original image, final SVM classifier is found aphid and is recorded Aphed population in MSER region;
Wherein, X, N, M are the natural number determined as required, and the value of M is greater than the value of N and the value of M, N is all less than X.
2. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: described small pixel subimage size is 30*20 pixel.
3. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: described wheat plant original image utilizes filtering operation to carry out pre-service to original image after obtaining, eliminate impurity effect.
4. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: described X value is that 5000-7000, N value is greater than 2000, M value and is greater than 5000.
5. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: described find in original image MSER utilize formula q (i)=| Qi+ △-Qi-△ |/| Qi| carries out, a certain connected region when Qi represents that threshold value is i, △ is the change of gray threshold, q (i) is the rate of change of region Qi when threshold value is i, when q (i) is for local minimum, then Qi is maximum stable extremal region.
6. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: described HOG characteristic extraction procedure is, by the original image gray processing of colour, utilize formula I (x, y)=I (x, y)
gammacarry out image standardization, wherein gamma gets 0.5, the gradient of each pixel in computed image, image is divided into several 6 × 6 pixel cell lattice and the histogram of gradients of adding up each cell forms the descriptor of each cell, every 3 × 3 cells are formed a block, it is exactly the HOG descriptor of this block that the descriptor of the cell in a block is together in series, and in image, the HOG Feature Descriptor of all pieces is together in series is exactly the HOG Feature Descriptor of this image.
7. according to a kind of field according to claim 6 wheat aphid method of counting, it is characterized in that: the gradient calculation method of described each pixel is, G
x(x, y)=H (x+1, y)-H (x-1, y), G
y(x, y)=H (x, y+1)-H (x, y-1), G
x(x, y), G
y(x, y), H (x, y) represents the horizontal direction gradient at input picture (x, y) place, vertical gradient and pixel value respectively,
,
, G (x, y) is gradient magnitude, and α (x, y) is gradient direction.
8. according to a kind of field according to claim 1 wheat aphid method of counting, it is characterized in that: after the positive and negative sample extraction of described HOG feature, carry out Haar feature detection.
9. according to a kind of field according to claim 8 wheat aphid method of counting, it is characterized in that: described Haar feature detection comprises, by edge feature, linear feature and central feature composition characteristic template, calculate Haar eigenwert by formula v=white pixel sum-2* black picture element sum.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510792807.8A CN105447859A (en) | 2015-11-18 | 2015-11-18 | Field wheat aphid counting method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510792807.8A CN105447859A (en) | 2015-11-18 | 2015-11-18 | Field wheat aphid counting method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105447859A true CN105447859A (en) | 2016-03-30 |
Family
ID=55557992
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510792807.8A Pending CN105447859A (en) | 2015-11-18 | 2015-11-18 | Field wheat aphid counting method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105447859A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106709437A (en) * | 2016-12-14 | 2017-05-24 | 北京工业大学 | Improved intelligent processing method for image-text information of scanning copy of early patent documents |
CN107688807A (en) * | 2016-08-05 | 2018-02-13 | 腾讯科技(深圳)有限公司 | Image processing method and image processing apparatus |
CN108052886A (en) * | 2017-12-05 | 2018-05-18 | 西北农林科技大学 | A kind of puccinia striiformis uredospore programming count method of counting |
CN108875825A (en) * | 2018-06-12 | 2018-11-23 | 河南工业大学 | A kind of Grainhouse injurious insect detection method based on image block |
CN109670398A (en) * | 2018-11-07 | 2019-04-23 | 北京农信互联科技集团有限公司 | Pig image analysis method and pig image analysis equipment |
CN109859161A (en) * | 2018-12-18 | 2019-06-07 | 青岛农业大学 | A kind of tubular type swells leg bee method of counting and its application |
CN109872301A (en) * | 2018-12-26 | 2019-06-11 | 浙江清华长三角研究院 | A kind of color image preprocess method counted for rice pest identification |
CN110659687A (en) * | 2019-09-24 | 2020-01-07 | 华南农业大学 | Yellow trapping plate-based phyllotreta striolata detection method, medium and equipment |
WO2020047738A1 (en) * | 2018-09-04 | 2020-03-12 | 安徽中科智能感知大数据产业技术研究院有限责任公司 | Automatic pest counting method based on combination of multi-scale feature fusion network and positioning model |
CN111160416A (en) * | 2019-12-12 | 2020-05-15 | 杭州电子科技大学 | Aedes mosquito identification method based on support vector machine |
CN117274812A (en) * | 2023-10-08 | 2023-12-22 | 北京香田智能科技有限公司 | Tobacco plant counting method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859382A (en) * | 2010-06-03 | 2010-10-13 | 复旦大学 | License plate detection and identification method based on maximum stable extremal region |
CN102096808A (en) * | 2011-01-19 | 2011-06-15 | 南京农业大学 | Method for automatically monitoring and reporting insect condition of rice planthopper |
US8213691B2 (en) * | 2008-12-30 | 2012-07-03 | Mitsubishi Electric Research Laboratories, Inc. | Method for identifying faces in images with improved accuracy using compressed feature vectors |
CN103049736A (en) * | 2011-10-17 | 2013-04-17 | 天津市亚安科技股份有限公司 | Face identification method based on maximum stable extremum area |
-
2015
- 2015-11-18 CN CN201510792807.8A patent/CN105447859A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8213691B2 (en) * | 2008-12-30 | 2012-07-03 | Mitsubishi Electric Research Laboratories, Inc. | Method for identifying faces in images with improved accuracy using compressed feature vectors |
CN101859382A (en) * | 2010-06-03 | 2010-10-13 | 复旦大学 | License plate detection and identification method based on maximum stable extremal region |
CN102096808A (en) * | 2011-01-19 | 2011-06-15 | 南京农业大学 | Method for automatically monitoring and reporting insect condition of rice planthopper |
CN103049736A (en) * | 2011-10-17 | 2013-04-17 | 天津市亚安科技股份有限公司 | Face identification method based on maximum stable extremum area |
Non-Patent Citations (3)
Title |
---|
YAO QING ETC: ""Automated Counting of Rice Planthoppers in Paddy Fields Based on Image Processing"", 《JOURNAL OF INTEGRATIVE AGRICULTURE》 * |
查宇飞 等: "《视频目标跟踪方法》", 31 July 2015 * |
王斌: ""基于MSER和PHOG特征的交通标志检测方法"", 《智慧工厂》 * |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107688807A (en) * | 2016-08-05 | 2018-02-13 | 腾讯科技(深圳)有限公司 | Image processing method and image processing apparatus |
CN106709437A (en) * | 2016-12-14 | 2017-05-24 | 北京工业大学 | Improved intelligent processing method for image-text information of scanning copy of early patent documents |
CN108052886B (en) * | 2017-12-05 | 2018-12-18 | 西北农林科技大学 | A kind of puccinia striiformis uredospore programming count method of counting |
CN108052886A (en) * | 2017-12-05 | 2018-05-18 | 西北农林科技大学 | A kind of puccinia striiformis uredospore programming count method of counting |
CN108875825B (en) * | 2018-06-12 | 2022-01-25 | 河南工业大学 | Granary pest detection method based on image blocking |
CN108875825A (en) * | 2018-06-12 | 2018-11-23 | 河南工业大学 | A kind of Grainhouse injurious insect detection method based on image block |
WO2020047738A1 (en) * | 2018-09-04 | 2020-03-12 | 安徽中科智能感知大数据产业技术研究院有限责任公司 | Automatic pest counting method based on combination of multi-scale feature fusion network and positioning model |
CN109670398A (en) * | 2018-11-07 | 2019-04-23 | 北京农信互联科技集团有限公司 | Pig image analysis method and pig image analysis equipment |
CN109859161A (en) * | 2018-12-18 | 2019-06-07 | 青岛农业大学 | A kind of tubular type swells leg bee method of counting and its application |
CN109872301A (en) * | 2018-12-26 | 2019-06-11 | 浙江清华长三角研究院 | A kind of color image preprocess method counted for rice pest identification |
CN109872301B (en) * | 2018-12-26 | 2022-07-15 | 浙江清华长三角研究院 | Color image preprocessing method for rice pest and disease identification and counting |
CN110659687A (en) * | 2019-09-24 | 2020-01-07 | 华南农业大学 | Yellow trapping plate-based phyllotreta striolata detection method, medium and equipment |
CN110659687B (en) * | 2019-09-24 | 2020-08-18 | 华南农业大学 | Yellow trapping plate-based phyllotreta striolata detection method, medium and equipment |
CN111160416A (en) * | 2019-12-12 | 2020-05-15 | 杭州电子科技大学 | Aedes mosquito identification method based on support vector machine |
CN111160416B (en) * | 2019-12-12 | 2023-03-21 | 杭州电子科技大学 | Aedes mosquito identification method based on support vector machine |
CN117274812A (en) * | 2023-10-08 | 2023-12-22 | 北京香田智能科技有限公司 | Tobacco plant counting method |
CN117274812B (en) * | 2023-10-08 | 2024-02-20 | 北京香田智能科技有限公司 | Tobacco plant counting method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105447859A (en) | Field wheat aphid counting method | |
Aquino et al. | A new methodology for estimating the grapevine-berry number per cluster using image analysis | |
CN104166841B (en) | The quick detection recognition methods of pedestrian or vehicle is specified in a kind of video surveillance network | |
CN109154978A (en) | System and method for detecting plant disease | |
CN103164692B (en) | A kind of special vehicle instrument automatic identification system based on computer vision and method | |
CN104680524B (en) | A kind of leafy vegetable disease screening method | |
US20150016687A1 (en) | Method, system and computer storage medium for face detection | |
CN105404867B (en) | A kind of substation isolating-switch state identification method of view-based access control model | |
CN109919002B (en) | Yellow stop line identification method and device, computer equipment and storage medium | |
CN107886507B (en) | A kind of salient region detecting method based on image background and spatial position | |
CN106156777A (en) | Textual image detection method and device | |
CN106203237A (en) | The recognition methods of container-trailer numbering and device | |
CN106446890A (en) | Candidate area extraction method based on window scoring and superpixel segmentation | |
CN106485273A (en) | A kind of method for detecting human face based on HOG feature and DNN grader | |
CN108109133A (en) | A kind of silkworm seed automatic counting method based on digital image processing techniques | |
Ye et al. | An image-based approach for automatic detecting tasseling stage of maize using spatio-temporal saliency | |
CN108830856B (en) | GA automatic segmentation method based on time series SD-OCT retina image | |
CN114049325A (en) | Construction method and application of lightweight face mask wearing detection model | |
Xuesong et al. | Aphid identification and counting based on smartphone and machine vision | |
CN109800616A (en) | A kind of two dimensional code positioning identification system based on characteristics of image | |
CN105354547A (en) | Pedestrian detection method in combination of texture and color features | |
CN105023272A (en) | Crop leaf insect pest detection method and system | |
CN110175650A (en) | A kind of power equipment automatic identifying method and device | |
CN111723656B (en) | Smog detection method and device based on YOLO v3 and self-optimization | |
CN113052234A (en) | Jade classification method based on image features and deep learning technology |
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: 20160330 |