CN110516686A - The mosquito recognition methods of three color RGB images - Google Patents

The mosquito recognition methods of three color RGB images Download PDF

Info

Publication number
CN110516686A
CN110516686A CN201910626708.0A CN201910626708A CN110516686A CN 110516686 A CN110516686 A CN 110516686A CN 201910626708 A CN201910626708 A CN 201910626708A CN 110516686 A CN110516686 A CN 110516686A
Authority
CN
China
Prior art keywords
mosquito
image
target
rgb
function
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.)
Granted
Application number
CN201910626708.0A
Other languages
Chinese (zh)
Other versions
CN110516686B (en
Inventor
谢雪梅
苗宏达
吴家骥
杨众杰
谭铭洲
李甫
付博勋
景易星
金星
韩笑
杜曜辛
李旭超
杨文哲
徐显梁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Electronic Science and Technology
Original Assignee
Xian University of Electronic Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xian University of Electronic Science and Technology filed Critical Xian University of Electronic Science and Technology
Priority to CN201910626708.0A priority Critical patent/CN110516686B/en
Publication of CN110516686A publication Critical patent/CN110516686A/en
Application granted granted Critical
Publication of CN110516686B publication Critical patent/CN110516686B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/25Determination of region of interest [ROI] or a volume of interest [VOI]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/50Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/56Extraction of image or video features relating to colour

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)

Abstract

The present invention discloses a kind of mosquito recognition methods of three color RGB images, mainly solves the prior art and needs computing resource more, the high problem of equipment cost.Its implementation is: obtaining the image-region of target mosquito, the pixel of traversal mosquito target image obtains the RGB color histogram and histogram functions of target mosquito image;According to three color RGB image histograms of variety classes mosquito target image, the RGB feature function of variety classes mosquito target image is designed;Function expression is carried out simplifying design according to the feature of specific mosquito specie and obtains the parameter of function;According to the RGB color histogram functions of target mosquito and mosquito RGB feature function, Classification and Identification is carried out to target mosquito.The present invention reduces computing resources, reduce equipment cost, are able to achieve the classification to two kinds of mosquitoes of Culex pipiens pallens and aedes albopictus, can be used for the identification of biological species.

Description

The mosquito recognition methods of three color RGB images
Technical field
The invention belongs to digital image processing techniques field, in particular to the mosquito in a kind of picture identifies, can be used for giving birth to The identification of object species or object.
Background technique
Traditional mosquito classification generally uses artificial method, even if with the naked eye to the Mosquito specimen being collected according to mosquito Morphological characteristic is classified, and it is low that there are classification effectivenesses, and risk is high, and by artificial characteristic, such as fatigue is affected The shortcomings that.The efficiency that mosquito identification can be greatlyd improve to mosquito identification using the method for Digital Image Processing, reduces manpower Cost and artificial risk.
With the development of computer and smart field, algorithm of target detection also achieves continuous breakthrough.Traditional target Detection algorithm is based primarily upon characteristics of image and is detected, and is to pass through if the Lowe scale invariant feature proposed converts SIFT algorithm It searches and is not easily susceptible to illumination, noise in image, the characteristic points such as affine transformation carry out the target in matching image, and use difference of Gaussian Function carries out extreme point detection, eliminates unstable extreme point, and the gradient of last use direction statistics with histogram key point obtains Specified clarification of objective.The SIFT algorithm has feature extraction good, the high feature of robustness.But there are algorithm complexity height, inspections The slow problem of degree of testing the speed.
The shortcomings that for SIFT, Yan Ke et al. and Bay et al. propose PCA-SIFT method and SURF method respectively. The method that PCA-SIFT algorithm introduces principal component analysis on the basis of SIFT, using PCA replace histogram to subvector into Row dimension-reduction treatment improves matching efficiency.SUFT introduces Hessian matrix to obtain crucial point location, considerably reduces meter Calculation amount.Paul Viola and Michael J Jones propose Viola-Jones algorithm be the first with universality can Handle and have in real time the Face datection algorithm of better effects.Viola-Jones algorithm uses Haar feature description graph piece window And the light and shade variation of regional area, operand is reduced using cascade classifier, realizes real-time target detection.
Existing target detection is mainly detected using deep learning, there is two class detection algorithms, and one kind is based on classification Algorithm of target detection, another kind of is that will classify to be converted into the algorithm of regression problem.Wherein the first kind mainly passes through OverFeat, R-CNN algorithm find the candidate region in image, classify to candidate region.Second class be mainly with YOLO and SSD is the algorithm of target detection of representative, they predict mesh target area in a manner of recurrence, and are measured with confidence level The probability of classification completes the detection and identification of target.
Since different mosquitoes have similar morphological characteristic, congener mosquito can show different postures, this There is certain difficulty for the characteristic point detection based on SIFT.And although the algorithm detection accuracy based on deep learning is very high, But a large amount of computing resource is needed, need more powerful calculating equipment, such as GPU equipment, therefore higher cost, Bu Nengman The demand of mosquito detection device is widely used in foot in larger region.
Summary of the invention
It is an object of the invention in view of the above shortcomings of the prior art, propose a kind of mosquito identification of three color RGB images Method reduces equipment cost to reduce computing resource, meets the needs of detecting in larger region to mosquito.
The technical scheme is that the characteristic being had differences by the background of picture to be detected and picture, extracts The mosquito target contained in picture to be detected;The color characteristic of mosquito target area is extracted, mosquito target area is obtained RGB color histogram.According to the statistical property design feature function of histogram and calculate the confidence score of variety classes mosquito. The identification situation of mosquito is obtained according to confidence score.Implementing step includes the following:
(1) single mosquito target image is obtained;
(2) traverse mosquito target image pixel obtain three color RGB image histograms of each mosquito target image with And thirdly color RGB image histogram functions H (i, c), wherein c ∈ (R, G, B) indicates the classification of three color RGB, i ∈ [1,256], Indicate the value of three color RGB;
(3) according to three color RGB image histograms of variety classes mosquito target image, variety classes mosquito target figure is designed The RGB feature function of picture:
Wherein [mosquito species] is mosquito specie set, and m is one kind in mosquito specie set, ScoremTable Show confidence level of the mosquito to be detected under the conditions of m class, function gm(i, c) and function fm(i, c) is illustrated respectively in right under the conditions of m class The two different form of weighting weight of the histogram functions h (i, c) of mosquito image to be detected;
(4) function expression is carried out simplifying design according to the feature of specific mosquito specie and obtains the parameter of function:
(4a) is arranged under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus, not to two kinds of histogram functions h (i, c) Weight is weighted with form:
Wherein, Setmc=[1, left] ∪ [right, 256] is a RGB value set, left SetmcThe right side of subset
Boundary, right SetmcThe left margin of subset;
(4b) set Score as mosquito to be detected setting under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus Reliability, by function fm(i, c), gm(i, c) substitutes into the RGB feature function in (3), the RGB feature function after being thus simplified Are as follows:
(4c) set the threshold value of confidence level of the mosquito to be detected under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus as Score is compared with thresh, differentiates to the type of mosquito to be detected by thresh: if Score > thresh, Then target mosquito is aedes albopictus, is Culex pipiens pallens otherwise;
The RGB histogram functions of Culex pipiens pallens and aedes albopictus image known to (4d) is one group given, traverse all Tri- kinds of parameters of left, right, thresh, in traversal each time, successively into RGB feature function, substitution one is given The value that the RGB histogram functions and left, right for the mosquito image known are traversed at this, the RGB feature function being calculated Value be Score, Score is compared with the current value of thresh, obtains the classification results of epicycle, every time traversal terminate Afterwards, it is quasi- to obtain mosquito classification results all after traversal for the accuracy rate that the classification results under parameter current are obtained by statistics The true highest parameter of rate is optimized parameter: left ', right ', thresh ';
(4e) obtains optimized parameter left ' obtained in (4d), right ' the RGB feature function being updated in (4b) most Whole characteristic function;
(5) it calculates target mosquito confidence level and classifies to target mosquito:
The histogram functions H (i, c) of the target mosquito image obtained in (2) is updated to spy final obtained in (4e) It levies in function, confidence of target mosquito image under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus is calculated Score ' is spent, which is compared with threshold value thresh ' obtained in (4d), the type of target mosquito is sentenced Other: if Score ' > thresh ', target mosquito is aedes albopictus, is Culex pipiens pallens otherwise.
Compared with the prior art, the present invention has the following advantages:
First, the present invention is extracted by the RGB property of the histogram to image object, and design feature function carries out mosquito The identification of worm is classified, and computation complexity is reduced, and the real-time inspection to mosquito target can be realized on the hardware device of low cost It surveys.
Second, the method for data characteristics is extracted by the present invention in that being used in given mosquito image data to calculate RGB The parameter of feature extraction function, and different types of target mosquito image data by detecting calculates different types of mosquito Confidence level, obtain the generic of target mosquito.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention.
Specific implementation measure
Example of the invention is described in further detail with reference to the accompanying drawing.
Referring to attached drawing 1, to the realization of example, steps are as follows:
Step 1. obtains single mosquito target image.
1.1) background picture is selected, background picture and target image to be detected are compared by the method for background subtraction, is obtained The distributed areas of mosquito target in image;
1.2) in the distributed areas of mosquito target, foreground image information is obtained by Background difference, passes through etching operation Picture noise is eliminated, the influence of mosquito texture is eliminated by expansive working, and then obtain mosquito distributed areas image;
1.3) connected region that each mosquito target is calculated separately on the image of mosquito distributed areas, obtains single mosquito mesh Mark area image.
Step 2. obtains RGB statistical property.
Mosquito target image pixel is traversed, three color RGB image distribution histogram H of mosquito target image are calculated (c, i),
Wherein, c ∈ (R, G, B) indicates the classification of three color RGB, and i ∈ [1,256] indicates the value of three color RGB, and R indicates red Color, G indicate green, and B indicates blue.
The RGB feature function of step 3. design variety classes mosquito target image.
3.1) the primitive character function of RGB is designed:
3.1.1 set [mosquito species] is set) as mosquito specie set, m is one kind in mosquito specie set, Histogram functions h (i, c) is obtained by the histogram of mosquito image to be detected;
3.1.2 it) according to the RGB distribution character of different types of mosquito, designs under the conditions of m class to mosquito image to be detected The two different form of weighting weight of histogram functions h (i, c) is respectively function gm(i, c) and function fm(i, c);
3.1.3) to obtain RGB primitive character function according to above-mentioned function as follows:
Wherein, ScoremIndicate confidence level of the mosquito to be detected under the conditions of m class;
3.2) primitive character function is simplified according to the feature of specific mosquito specie:
It is arranged under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus, to two kinds of histogram functions h (i, c) not similar shapes It is as follows that formula weights weight difference:
fm(i, c)=1
Wherein, Setmc=[1, left] ∪ [right, 256] is a RGB value set, left SetmcThe right side of subset Boundary, right SetmcThe left margin of subset;
3.3) it is distributed by having the RGB of mosquito data, the parameter of function is obtained using statistical law.
3.3.1 Score is set) as mosquito to be detected under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus Confidence level, if the threshold value of confidence level of the mosquito to be detected under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus is thresh; Score is compared with thresh, the type of mosquito to be detected is differentiated: if Score > thresh, target mosquito Worm is aedes albopictus, is Culex pipiens pallens otherwise;
3.3.2) give one group known to Culex pipiens pallens and aedes albopictus image RGB histogram functions, traverse all RGB value set SetmcBoth parameters of the right margin left of subset, left margin right;And in traversal nesting to be detected The traversal of confidence threshold value thresh of mosquito under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus, obtains in 0 to 1 range Traversal value thresh ' that is interior, being 0.01 with step-length;
3.3.3 a given known mosquito figure) is successively substituted into RGB feature function in the traversal to thresh Value left ', the right ' that the RGB histogram functions and right boundary left, right of picture are traversed at this, are calculated RGB The value of characteristic function is Score ', and the Score ' and thresh ' is compared, and records the accurate of the classification results of epicycle Rate, after the traversal to thresh, the corresponding parameter of accuracy rate that highest classification results are obtained by counting ranking is Left ", right ", thresh ";Change left, the value of right carries out next round traversal, and to left, the traversal of right is whole After, accuracy rate obtained in aforementioned traversal is compared, obtains mosquito classification results accuracy rate most by counting ranking High parameter is optimized parameter: left " ', right " ', thresh " ';
3.4) by the function g in 3.2)m(i, c) and function fm(i, c) and 3.3.3) obtained in optimized parameter left " ', Right " ' is updated to 3.1.3) in RGB primitive character function obtain final characteristic function:
Wherein
Step 4. calculates mosquito confidence level and classifies to target mosquito.
The histogram functions H (i, c) of the target mosquito image obtained in step 2 is updated to obtained in step 3.4) most In whole characteristic function, target mosquito image is calculated under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus Confidence level Score ", the Score " value and 3.3.3) obtained in threshold value thresh " ' is compared, to target mosquito Type is differentiated: if Score " > thresh " ', target mosquito is aedes albopictus, is Culex pipiens pallens otherwise.
Above description is only example of the present invention, does not constitute any limitation of the invention, it is clear that for It, all may be without departing substantially from the principle of the invention, structure after having understood the content of present invention and principle for one of skill in the art In the case where, carry out various modifications and change in form and details, but these modifications and variations based on inventive concept Still within the scope of the claims of the present invention.

Claims (2)

1. a kind of mosquito recognition methods of three color RGB images, which is characterized in that include the following:
(1) single mosquito target image is obtained;
(2) traverse mosquito target image pixel obtain each mosquito target image three color RGB image histograms and its Three color RGB image histogram functions H (i, c), wherein c ∈ (R, G, B), indicates the classification of three color RGB, and i ∈ [1,256] is indicated The value of three color RGB;
(3) according to three color RGB image histograms of variety classes mosquito target image, variety classes mosquito target image is designed RGB feature function:
Wherein [mosquito species] is mosquito specie set, and m is one kind in mosquito specie set, ScoremIndicate to Detect confidence level of mosquito under the conditions of m class, function gm(i, c) and function fm(i, c) is illustrated respectively under the conditions of m class to be checked Survey the two different form of weighting weight of the histogram functions h (i, c) of mosquito image;
(4) function expression is carried out simplifying design according to the feature of specific mosquito specie and obtains the parameter of function:
(4a) is arranged under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus, to two kinds of histogram functions h (i, c) not similar shapes Formula weights weight:
fm(i, c)=1,
Wherein, Setmc=[1, left] ∪ [right, 256] is a RGB value set, left SetmcThe right margin of subset, Right is SetmcThe left margin of subset;
(4b) sets confidence of the Score as mosquito to be detected under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus Degree, by function fm(i, c), gm(i, c) substitutes into the RGB feature function in (3), the RGB feature function after being thus simplified are as follows:
(4c) sets the threshold value of confidence level of the mosquito to be detected under the conditions of two class mosquito of Culex pipiens pallens and aedes albopictus as thresh, Score is compared with thresh, the type of mosquito to be detected is differentiated: if Score > thresh, target mosquito Worm is aedes albopictus, is Culex pipiens pallens otherwise;
The RGB histogram functions of Culex pipiens pallens and aedes albopictus image known to (4d) is one group given, traverse all left, Tri- kinds of parameters of right, thresh successively substitute into a given known mosquito in traversal each time into RGB feature function The RGB histogram functions and left of worm image, the value of the value that right is traversed at this, the RGB feature function being calculated are Score is compared with the current value of thresh, obtains the classification results of epicycle by Score, every time after traversal, warp The accuracy rate for the classification results that statistics obtains under parameter current is crossed, all after traversal, obtains mosquito classification results accuracy rate Highest parameter is optimized parameter: left " ', right " ', thresh " ';
(4e) obtains the RGB feature function for optimized parameter left " ', right " obtained in (4d) ' be updated in (4b) finally Characteristic function;
(5) it calculates target mosquito confidence level and classifies to target mosquito:
The histogram functions H (i, c) of the target mosquito image obtained in (2) is updated to feature letter final obtained in (4e) In number, confidence level of target mosquito image under conditions of just for two class mosquito of Culex pipiens pallens and aedes albopictus is calculated The Score " value and threshold value thresh obtained in (4d) " ' is compared, sentences to the type of target mosquito by Score " Other: if Score " > thresh " ', target mosquito is aedes albopictus, is Culex pipiens pallens otherwise.
2., wherein obtaining single mosquito target image in (1), being accomplished by according to claim and the method
(1a) compares background picture and target image to be detected by the method for background subtraction, obtains mosquito target in image Distributed areas;Foreground image information is obtained by Background difference, picture noise is eliminated by etching operation, is disappeared by expansive working Except the influence of mosquito texture, and then obtain mosquito distributed areas image;
(1b) calculates separately the connected region of each mosquito target on the image of mosquito distributed areas, obtains single mosquito target area Area image.
CN201910626708.0A 2019-07-11 2019-07-11 Mosquito recognition method of three-color RGB image Active CN110516686B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910626708.0A CN110516686B (en) 2019-07-11 2019-07-11 Mosquito recognition method of three-color RGB image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910626708.0A CN110516686B (en) 2019-07-11 2019-07-11 Mosquito recognition method of three-color RGB image

Publications (2)

Publication Number Publication Date
CN110516686A true CN110516686A (en) 2019-11-29
CN110516686B CN110516686B (en) 2021-03-23

Family

ID=68622464

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910626708.0A Active CN110516686B (en) 2019-07-11 2019-07-11 Mosquito recognition method of three-color RGB image

Country Status (1)

Country Link
CN (1) CN110516686B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102915446A (en) * 2012-09-20 2013-02-06 复旦大学 Plant disease and pest detection method based on SVM (support vector machine) learning
US9953242B1 (en) * 2015-12-21 2018-04-24 Amazon Technologies, Inc. Identifying items in images using regions-of-interest
US10007863B1 (en) * 2015-06-05 2018-06-26 Gracenote, Inc. Logo recognition in images and videos
CN108280483A (en) * 2018-01-30 2018-07-13 华南农业大学 Trypetid adult image-recognizing method based on neural network
CN108647665A (en) * 2018-05-18 2018-10-12 西安电子科技大学 Vehicle real-time detection method of taking photo by plane based on deep learning
CN109492581A (en) * 2018-11-09 2019-03-19 中国石油大学(华东) A kind of human motion recognition method based on TP-STG frame

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102915446A (en) * 2012-09-20 2013-02-06 复旦大学 Plant disease and pest detection method based on SVM (support vector machine) learning
US10007863B1 (en) * 2015-06-05 2018-06-26 Gracenote, Inc. Logo recognition in images and videos
US9953242B1 (en) * 2015-12-21 2018-04-24 Amazon Technologies, Inc. Identifying items in images using regions-of-interest
CN108280483A (en) * 2018-01-30 2018-07-13 华南农业大学 Trypetid adult image-recognizing method based on neural network
CN108647665A (en) * 2018-05-18 2018-10-12 西安电子科技大学 Vehicle real-time detection method of taking photo by plane based on deep learning
CN109492581A (en) * 2018-11-09 2019-03-19 中国石油大学(华东) A kind of human motion recognition method based on TP-STG frame

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
MONA MINAKSHI 等: "Identifying mosquito species using smart-phone cameras", 《2017 EUROPEAN CONFERENCE ON NETWORKS AND COMMUNICATIONS (EUCNC)》 *
徐源浩 等: "基于ELM理论的昆虫分类", 《电子科技》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Also Published As

Publication number Publication date
CN110516686B (en) 2021-03-23

Similar Documents

Publication Publication Date Title
Zhao et al. Cloud shape classification system based on multi-channel cnn and improved fdm
Indriani et al. Tomatoes classification using K-NN based on GLCM and HSV color space
CN109154978B (en) System and method for detecting plant diseases
Shi et al. Automatic road crack detection using random structured forests
CN109145921B (en) Image segmentation method based on improved intuitive fuzzy C-means clustering
CN106295124B (en) The method of a variety of image detecting technique comprehensive analysis gene subgraph likelihood probability amounts
Al-Hiary et al. Fast and accurate detection and classification of plant diseases
Chung et al. Efficient shadow detection of color aerial images based on successive thresholding scheme
Al Bashish et al. Detection and classification of leaf diseases using K-means-based segmentation and
KR101964397B1 (en) Information processing apparatus and information processing method
CN108932479A (en) A kind of human body anomaly detection method
CN105718866B (en) A kind of detection of sensation target and recognition methods
CN108288033A (en) A kind of safety cap detection method merging multiple features based on random fern
Trivedi et al. Automatic segmentation of plant leaves disease using min-max hue histogram and k-mean clustering
Masood et al. Plants disease segmentation using image processing
CN115994907B (en) Intelligent processing system and method for comprehensive information of food detection mechanism
CN112257702A (en) Crop disease identification method based on incremental learning
Kumar et al. Multiclass support vector machine based plant leaf diseases identification from color, texture and shape features
CN109145770B (en) Automatic wheat spider counting method based on combination of multi-scale feature fusion network and positioning model
Hernández et al. Assessment of downy mildew in grapevine using computer vision and fuzzy logic. Development and validation of a new method
Valkonen et al. Dual structured convolutional neural network with feature augmentation for quantitative characterization of tissue histology
CN114037650A (en) Ground target visible light damage image processing method for change detection and target detection
CN110516686A (en) The mosquito recognition methods of three color RGB images
Zeng et al. Detecting and measuring fine roots in minirhizotron images using matched filtering and local entropy thresholding
CN116416523A (en) Machine learning-based rice growth stage identification system and method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant