CN106157266A - A kind of orchard fruit image acquiring method - Google Patents
A kind of orchard fruit image acquiring method Download PDFInfo
- Publication number
- CN106157266A CN106157266A CN201610532804.5A CN201610532804A CN106157266A CN 106157266 A CN106157266 A CN 106157266A CN 201610532804 A CN201610532804 A CN 201610532804A CN 106157266 A CN106157266 A CN 106157266A
- Authority
- CN
- China
- Prior art keywords
- image
- fruit
- pixel
- significance
- super
- 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
- 235000013399 edible fruits Nutrition 0.000 title claims abstract description 58
- 238000000034 method Methods 0.000 title claims abstract description 26
- 239000002420 orchard Substances 0.000 title claims abstract description 13
- 238000001514 detection method Methods 0.000 claims abstract description 25
- 230000011218 segmentation Effects 0.000 claims abstract description 9
- 238000005728 strengthening Methods 0.000 claims abstract description 4
- 238000001914 filtration Methods 0.000 claims description 12
- 230000003044 adaptive effect Effects 0.000 claims description 7
- 241000238876 Acari Species 0.000 claims description 3
- 238000007596 consolidation process Methods 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 2
- 235000007270 Gaultheria hispida Nutrition 0.000 description 6
- 235000009134 Myrica cerifera Nutrition 0.000 description 6
- 244000024215 Myrica gale Species 0.000 description 6
- 235000012851 Myrica pensylvanica Nutrition 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 4
- 238000005286 illumination Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000003306 harvesting Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000001124 body fluid Anatomy 0.000 description 1
- 239000010839 body fluid Substances 0.000 description 1
- 235000009508 confectionery Nutrition 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000029087 digestion Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004345 fruit ripening Effects 0.000 description 1
- 238000000227 grinding Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000035922 thirst Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- 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
- G06T2207/20004—Adaptive image processing
-
- 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
- G06T2207/20048—Transform domain processing
- G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of orchard fruit image acquiring method, comprise the following steps: (1) image acquisition step: view-based access control model sensor Real-time Collection fruit image；(2) image enhancement step: the image gathered in step (1) is carried out image enhaucament, while strengthening the image detail of dark space, does not the most lose the image detail in clear zone, reduces natural lighting impact；(3) significance target detection step: image after strengthening in step (2) is carried out super-pixel segmentation, builds closed loop graph model, obtain fruit significance target based on two-stage significance target detection model；(4) fruit image acquisition step: the notable figure of fruit detected in step (3) is split by Otsu method, it is thus achieved that two-value black white image, then superpose with original image, thus fruit is split from background.The present invention need not use the feature such as fruit color or shape, has versatility, can promote the practicalization of fruit picking robot.
Description
Technical field
The invention belongs to technical field of image processing, relate to a kind of orchard fruit image acquiring method, especially natural ring
The acquisition of the fruit image of uneven illumination is there is under border.
Background technology
Fructus Myricae rubrae is Chinese tradition local product fruits, and fruit sweet and sour is moderate, direct-edible, has simultaneously and quenches the thirst, promotes the production of body fluid, helps
The functions such as digestion, have the highest medicinal and edibility.And Fructus Myricae rubrae fruit tree is tall and big, it is intensive to bear fruit, the fruit maturation phase is shorter, and
Being affected by plum rains hot weather, easily rot or come off, reduce its commodity value, orchard worker's loss is the most serious.Therefore fruit
Real timely collecting, reduction harvesting cost are particularly important.Fructus Myricae rubrae picking robot can save labour force, it is achieved rapid picking, carries
High Fructus Myricae rubrae plucks quality, thus promotes economic benefit, increases orchard worker's income.
Along with the development of modern agriculture, picking robot based on machine vision becomes grinding of domestic and international agricultural engineering field
Study carefully focus.And the acquisition to fruit image is picking robot succeeding target identification location, the top priority of harvesting.
Summary of the invention
It is an object of the invention to: provide a kind of orchard fruit image acquiring method so that fruit picking robot is at image
Processing stage be capable of the accurate acquisition to fruit image, can complete further to identify location, promote fruit picking robot
Practicalization.Realize technical scheme to comprise the steps:
(1) image acquisition step: view-based access control model sensor Real-time Collection fruit image.
(2) image enhancement step: this step uses homomorphic filtering algorithm for image enhancement, enters the image of uneven illumination
Row light compensation, makes whole pixel distribution more they tend to rationally, while strengthening picture contrast, strengthens image dark place details,
So that fruit is the most notable clear in image.
(3) significance target detection step: first, uses SLIC super-pixel partitioning algorithm, fruit image is divided into shape
Shape rule, the uniform super-pixel of in the same size and vision, to reduce the quantity of image process target, improves the effect of subsequent treatment
Rate；Build non-directed graph model with super-pixel as node, to shorten the geodesic distance between similar super-pixel, improve follow-up significantly
Property detection accuracy；Then the closed loop graph model of structure is modeled in image data set manifold structure；Then,
In one-level well-marked target detection model, first manifold ranking algorithm is passed through separately as query node in each border on four borders
It is ranked up, it is thus achieved that four different significance testing results, uses the consolidation strategy of multiplication to be closed by these four notable figures
And, draw the significance target that the first order detects；Finally, in the well-marked target detection model of the second level, the equal of notable figure is chosen
Value, as adaptive threshold, carries out binary segmentation to the notable figure of the first order, and the vertex ticks that will be greater than threshold value is target query
Node, and build a label vector y accordingly, then obtain ordering vector f with the algorithm of manifold ranking*, by normalized sequence
Value is defined as the saliency value of super-pixel node, it is thus achieved that the second level is significantly schemed.
(4) fruit image acquisition step: this step is that figure notable to the second level obtained carries out adaptive by Otsu method
Answer Threshold segmentation, it is thus achieved that two-value black white image, then superpose with original image, thus fruit is split from background.
Beneficial effects of the present invention:
(1) for fruit picking robot, this inventive method is capable of the segmentation of Waxberry fruit image, has obtained
Whole fruit region, plays an important role to promoting the practical of fruit picking robot.
(2) before fruit image acquisition step, the detection of Waxberry fruit significance target has been carried out, it is possible to need not basis
The difference of the color of Waxberry fruit or shape facility etc. and background is split, thus has certain versatility, is applicable to
Other field grown fruits similar with Fructus Myricae rubrae.
(3) in image enhancement step, algorithm for image enhancement uses Homomorphic Filtering Algorithm, it is possible to compensate light, enhancing contrast ratio
While, dark place details have also been obtained enhancing, makes fruit the most notable clear.
(4) the closed loop graph model being built into is modeled to by significance target detection step in image data set manifold
Structure, carries out the assignment of query point significance value, it is thus achieved that Saliency maps by manifold ranking algorithm.
(5) significance target detection step have employed two-stage significance target detection model, by manifold ranking algorithm
Two grades of sequences, inhibit some singular points present in background well.
(6) fruit image acquisition step uses the notable figure in the Otsu method second level to obtaining carry out adaptive threshold to divide
Cut, then superpose with original image, thus fruit is split from background.
Accompanying drawing explanation
Fig. 1 is that fruit image splits main-process stream；
Fig. 2 is homomorphic filtering image enhaucament flow process；
Fig. 3 is significance target detection flow process；
Fig. 4 is that Waxberry fruit obtains design sketch.
Detailed description of the invention
Below in conjunction with the accompanying drawings embodiments of the present invention are further described.The present invention says as a example by Fructus Myricae rubrae
Bright, but present disclosure applies equally to other fruits similar.
As it is shown in figure 1, the orchard Waxberry fruit image acquiring method that the present invention proposes comprises the steps:
(1) image acquisition step
The collection view-based access control model sensor of image, as the area information of subsequent extracted destination object, gathers image such as Fig. 4
Shown in (a).
(2) image enhancement step
This step implementing procedure is as shown in Figure 2.In view of there is the phenomenon of uneven illumination under natural environment, can affect notable
Property target detection, and then affect image segmentation effect, use homomorphic filtering algorithm for image enhancement carry out light compensation.Should
Step uses homomorphic filtering algorithm for image enhancement, empty by collection image is transformed into hsv color from RGB color
Between, carry out fast Fourier transform (FFT) for brightness V component after taking the logarithm, available frequency-domain expression, select dynamic Bart
(u, v) carries out frequency domain filtering to Butterworth homomorphic filtering function H, and formula (1) is shown in function definition, then carries out Inverse Fast Fourier Transforms
(IFFT) and exponential transform, so obtain filtering enhanced image such as Fig. 4 (b).
R in formula (1)LRepresent low-frequency gain, 0 < rL<1；rHRepresent high-frequency gain, rH>1；Constant c is at rLAnd rHBetween transition, use
Control the steepness on filter function inclined-plane；M, n are Dynamic Operator；D (u, v) be point (u, v) to Fourier transform center away from
From, see formula (2):
In formula (2), the line number of M, N representative image respectively and columns.
(3) significance target detection step
This step implementing procedure is as shown in Figure 3.Extraction step in image is gathered for Fructus Myricae rubrae include:
First, utilize SLIC super-pixel to produce algorithm and carry out super-pixel segmentation, therebetween, by pixel-map to LABXY space,
Being expressed as the eigen vector [l, a, b, x, y] of five dimensions, wherein, [l, a, b] is the pixel color characteristic at CIElab color space,
[x, y] is the space coordinates of pixel, by two pixels i of definition, the distance between j, sees formula (3), (4), (5), utilizes K-
Pixel is clustered in the range of local by means means clustering algorithm, obtains regular shape, in the same size and vision is uniform
Super-pixel, such as Fig. 4 (c).
In formula (3), m is a constant in the range of [1,20], is used for controlling the compactness of super-pixel, weighs color value
With spatial information proportion in measuring similarity,N is the total number of pixels of image, and K is a super-pixel to be generated
Number.
Then, build the non-directed graph model with super-pixel as node, be the most merely to have between the two in non-directed graph model
It is connected between the node of common boundary, also includes being connected between node conterminal with adjacent node, and by the four of image
Individual border is connected, thus constitutes a closed loop graph model.
Then, the closed loop graph model being built into is modeled in image data set manifold structure, use two-stage notable
Property target detection model carries out Myricales target detection.In first order well-marked target detection model, first every by four borders
Individual border, separately as query node, is iterated by the iterative equation of manifold ranking algorithm, sees formula (6), obtains all joints
The ordering vector f of point*, see formula (7), by f*Normalize to the ordering vector f after [0,1] scope-*, define super by formula (8)
The significance S of pixel node ibI (), it is thus achieved that be four different significance object detection results, uses the consolidation strategy of multiplication
These four notable figures are merged, draws the significance target that the first order detects, such as Fig. 4 (d).
F (t+1)=α Sf (t)+(1-α) y (6)
f*=(1-α) (D-α W)-1y (7)
Sb(i)=1-f-*(i) i=1,2 ..., N (8)
Wherein: α is expressed as initial ranking value and neighborhood propagates the contribution degree to final ranking value, W=[wij]n×nIt is expressed as
The similarity matrix of figure, corresponding to the weight of limit E,wij∈ W, ciAnd cjRepresent super-pixel joint respectively
Point i and j is at the average of CIELab color space, σwRepresent the intensity of control weight；D=diag{d11,...,dnnIt is expressed as figure
Diagonal matrix, dii=∑jwij；S=D-1/2WD-1/2It is expressed as the normalization matrix of the similarity matrix of figure；f:X→RnRepresent
It is a sequence equation, it is possible to see a vector f=[f as1,...,fn]T；Y=[y1,...,yn]TRepresent a label to
Amount, works as xiFor y during query nodei=1, otherwise yi=0, data set X={x1,...,xl,xl+1,...,xn}∈Rn×n, correspond to
Non-directed graph model G (V, E), the node in V representative graph, the unmarked joint being ranked up including marked query node and needs
Point, the nonoriented edge in E representative graph.
In the well-marked target detection model of the second level, choose the average of notable figure as adaptive threshold, to the first order
Notable figure carries out binary segmentation, and the vertex ticks that will be greater than threshold value is target query node, and builds a label vector accordingly
Y, then obtain ordering vector f with the algorithm of manifold ranking*, normalized ranking value is defined as the saliency value of super-pixel node,
See formula (9), detect that the second level is significantly schemed, obtain Waxberry fruit significance target with this, such as Fig. 4 (e).
Sfq(i)=f-*(i) i=1,2 ..., N (9)
(4) fruit image acquisition step
This step is to use the notable figure in the Otsu method second level to obtaining to carry out adaptive threshold fuzziness, obtains two-value black
White image, then superpose with original image, thus fruit is split from background, as shown in Fig. 4 (f).
Embodiment of above is merely to illustrate technical scheme, and not limitation of the present invention, relevant technology
The those of ordinary skill in field, without departing from the spirit and scope of the present invention, it is also possible to make a variety of changes, therefore
The technical scheme of all equivalents falls within the category of present invention protection.
Claims (5)
1. an orchard fruit image acquiring method, it is characterised in that comprise the following steps:
(1) image acquisition step: view-based access control model sensor Real-time Collection fruit image；
(2) image enhancement step: the image gathered in step (1) is carried out image enhaucament, strengthens image detail same of dark space
Time, the most do not lose the image detail in clear zone, reduce natural lighting impact；
(3) significance target detection step: image after strengthening in step (2) is carried out super-pixel segmentation, builds closed loop graph model,
Fruit significance target is obtained based on two-stage significance target detection model；
(4) fruit image acquisition step: the notable figure of fruit detected in step (3) is split, it is thus achieved that two-value black and white
Image, then superpose with original image, thus fruit is split from background.
A kind of orchard the most according to claim 1 fruit image acquiring method, it is characterised in that the tool of described step (2)
Body implementation method includes: use homomorphic filtering algorithm for image enhancement, by being transformed into from RGB color by the image of collection
Hsv color space, carries out fast Fourier transform, obtains frequency-domain expression, select dynamic bar after taking the logarithm for brightness V component
(u, v) carries out frequency domain filtering to special Butterworth homomorphic filtering function H, then carries out Inverse Fast Fourier Transforms and exponential transform, is filtered
The enhanced image of ripple.
A kind of orchard the most according to claim 2 fruit image acquiring method, it is characterised in that described dynamic Butterworth
Homomorphic filtering function H (u, expression formula v) is as follows:
R in formulaLRepresent low-frequency gain, 0 < rL<1；rHRepresent high-frequency gain, rH>1；Constant c is at rLAnd rHBetween transition, be used for control
The steepness on filter function inclined-plane；M, n are Dynamic Operator；(u is v) that (u, v) to the distance of Fourier transform center for point to D；The line number of M, N representative image respectively and columns.
A kind of orchard the most according to claim 1 fruit image acquiring method, it is characterised in that the tool of described step (3)
Body implementation method comprises the following steps:
(3-1) utilize SLIC super-pixel to produce algorithm and carry out super-pixel segmentation, therebetween, by pixel-map to LABXY space, represent
Being the eigen vector [l, a, b, x, y] of five dimensions, wherein, [l, a, b] is the pixel color characteristic at CIElab color space, [x,
Y] it is the space coordinates of pixel；By definition distance between two pixels i, j, utilize K-means means clustering algorithm in office
In the range of portion, pixel is clustered, obtain regular shape, the uniform super-pixel of in the same size and vision；
(3-2) build the non-directed graph model with super-pixel as node, be the most merely to have between the two jointly in non-directed graph model
It is connected between the node on border, also includes being connected between node conterminal with adjacent node, and by the four of image limits
Boundary is connected, thus constitutes a closed loop graph model；
(3-3) the closed loop graph model being built into is modeled in image data set manifold structure, use two-stage significance mesh
Mark detection model carries out the detection of fruit object:
In first order well-marked target detection model, first by each border on four borders separately as query node, by stream
The iterative equation of shape sort algorithm is iterated, and then obtains the ordering vector f of all nodes*, by f*Normalize to [0,1] model
The ordering vector f enclosed-*, by expression formula Sb(i)=1-f-*(i) i=1,2 ..., N defines the significance of super-pixel node i
SbI (), it is thus achieved that be four different significance object detection results, uses the consolidation strategy of multiplication to be carried out by these four notable figures
Merge, draw the significance target that the first order detects；
In the well-marked target detection model of the second level, choose the average of notable figure as adaptive threshold, notable to the first order
Figure carries out binary segmentation, and the vertex ticks that will be greater than threshold value is target query node, and builds a label vector y accordingly, then
Ordering vector f is obtained with the algorithm of manifold ranking*, normalized ranking value is defined as the saliency value of super-pixel node,
Detect that the second level is significantly schemed, obtain fruit significance target with this.
A kind of orchard the most according to claim 1 fruit image acquiring method, it is characterised in that the tool of described step (4)
Body implementation method includes: use the notable figure in the Otsu method second level to obtaining to carry out adaptive threshold fuzziness, it is thus achieved that two-value is black
White image, then superpose with original image, thus fruit is split from background.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610532804.5A CN106157266A (en) | 2016-07-07 | 2016-07-07 | A kind of orchard fruit image acquiring method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610532804.5A CN106157266A (en) | 2016-07-07 | 2016-07-07 | A kind of orchard fruit image acquiring method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106157266A true CN106157266A (en) | 2016-11-23 |
Family
ID=58062899
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610532804.5A Pending CN106157266A (en) | 2016-07-07 | 2016-07-07 | A kind of orchard fruit image acquiring method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106157266A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815843A (en) * | 2016-11-30 | 2017-06-09 | 江苏城乡建设职业学院 | A kind of fruit object acquisition methods based on convex closure center priori and absorbing Marcov chain |
| CN108537744A (en) * | 2018-03-14 | 2018-09-14 | 北京师范大学 | A kind of coloured image luminance component homomorphic filtering defogging method |
| CN108711140A (en) * | 2018-05-16 | 2018-10-26 | 广东欧谱曼迪科技有限公司 | A kind of image brightness uniformity real-time recovery method based on inter-class variance description |
| CN109583455A (en) * | 2018-11-20 | 2019-04-05 | 黄山学院 | A kind of image significance detection method merging progressive figure sequence |
| CN110136161A (en) * | 2019-05-31 | 2019-08-16 | 苏州精观医疗科技有限公司 | Image characteristics extraction analysis method, system and device |
| CN113469976A (en) * | 2021-07-06 | 2021-10-01 | 浙江大华技术股份有限公司 | Object detection method and device and electronic equipment |
| CN113506312A (en) * | 2021-06-24 | 2021-10-15 | 上海电力大学 | Ultraviolet discharge image segmentation method and computer readable medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2383680A1 (en) * | 2010-04-29 | 2011-11-02 | Dralle A/S | Classification of objects in harvesting applications |
| CN103745468A (en) * | 2014-01-07 | 2014-04-23 | 上海交通大学 | Significant object detecting method based on graph structure and boundary apriority |
| CN105701829A (en) * | 2016-01-16 | 2016-06-22 | 常州大学 | Bagged green fruit image segmentation method |
-
2016
- 2016-07-07 CN CN201610532804.5A patent/CN106157266A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2383680A1 (en) * | 2010-04-29 | 2011-11-02 | Dralle A/S | Classification of objects in harvesting applications |
| CN103745468A (en) * | 2014-01-07 | 2014-04-23 | 上海交通大学 | Significant object detecting method based on graph structure and boundary apriority |
| CN105701829A (en) * | 2016-01-16 | 2016-06-22 | 常州大学 | Bagged green fruit image segmentation method |
Non-Patent Citations (5)
| Title |
|---|
| CHUAN YANG等: "Saliency Detection via Graph-Based Manifold Ranking", 《COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2013 IEEE CONFERENCE ON》 * |
| RADHAKRISHNA ACHANTA等: "SLIC Superpixels", 《TECHNICAL REPORT, EPFL》 * |
| SHUZHI SAM GE等: "Bottom-up saliency detection for attention determination", 《MACHINE VISION AND APPLICATIONS》 * |
| 徐黎明等: "自然环境下杨梅果实图像的分割方法研究", 《沈阳农业大学学报》 * |
| 陈科尹等: "基于视觉显著性改进的水果图像模糊聚类分割算法", 《农业工程学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815843A (en) * | 2016-11-30 | 2017-06-09 | 江苏城乡建设职业学院 | A kind of fruit object acquisition methods based on convex closure center priori and absorbing Marcov chain |
| CN108537744A (en) * | 2018-03-14 | 2018-09-14 | 北京师范大学 | A kind of coloured image luminance component homomorphic filtering defogging method |
| CN108711140A (en) * | 2018-05-16 | 2018-10-26 | 广东欧谱曼迪科技有限公司 | A kind of image brightness uniformity real-time recovery method based on inter-class variance description |
| CN108711140B (en) * | 2018-05-16 | 2021-09-10 | 广东欧谱曼迪科技有限公司 | Image brightness uniformity real-time recovery method based on inter-class variance description |
| CN109583455A (en) * | 2018-11-20 | 2019-04-05 | 黄山学院 | A kind of image significance detection method merging progressive figure sequence |
| CN110136161A (en) * | 2019-05-31 | 2019-08-16 | 苏州精观医疗科技有限公司 | Image characteristics extraction analysis method, system and device |
| CN113506312A (en) * | 2021-06-24 | 2021-10-15 | 上海电力大学 | Ultraviolet discharge image segmentation method and computer readable medium |
| CN113469976A (en) * | 2021-07-06 | 2021-10-01 | 浙江大华技术股份有限公司 | Object detection method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106157266A (en) | A kind of orchard fruit image acquiring method | |
| CN103996018B (en) | Face identification method based on 4DLBP | |
| CN105844285A (en) | Cucumber disease identification method and apparatus based on image information | |
| Sunny et al. | An efficient citrus canker detection method based on contrast limited adaptive histogram equalization enhancement | |
| CN106845497B (en) | Corn early-stage image drought identification method based on multi-feature fusion | |
| CN101551853A (en) | Human ear detection method under complex static color background | |
| Sun et al. | Recognition of green apples based on fuzzy set theory and manifold ranking algorithm | |
| CN106294705A (en) | A kind of batch remote sensing image preprocess method | |
| CN111223110B (en) | Microscopic image enhancement method and device and computer equipment | |
| Ullagaddi et al. | Disease recognition in Mango crop using modified rotational kernel transform features | |
| CN111798470B (en) | Crop image entity segmentation method and system applied to intelligent agriculture | |
| Ji et al. | In-field automatic detection of maize tassels using computer vision | |
| Revathi et al. | Homogenous segmentation based edge detection techniques for proficient identification of the cotton leaf spot diseases | |
| Patel et al. | A survey on the plant leaf disease detection techniques | |
| CN110276759A (en) | A kind of bad line defect diagnostic method of Mobile phone screen based on machine vision | |
| CN114758132B (en) | Fruit tree disease and pest identification method and system based on convolutional neural network | |
| Liu et al. | Recognition and localization of actinidia arguta based on image recognition | |
| CN110516648B (en) | Ramie plant number identification method based on unmanned aerial vehicle remote sensing and pattern identification | |
| CN106529535A (en) | License plate positioning method in complex lighting environment based on wavelet transform | |
| Singh et al. | Automatic detection of rust disease of Lentil by machine learning system using microscopic images. | |
| CN112329506A (en) | Fruit identification method and system, and positioning method and system of wolfberry picking robot | |
| CN111563542A (en) | Automatic plant classification method based on convolutional neural network | |
| CN114596274A (en) | Natural background citrus greening disease detection method based on improved Cascade RCNN network | |
| Wang et al. | A fast image segmentation algorithm for detection of pseudo-foreign fibers in lint cotton | |
| CN105631451A (en) | Plant leave identification method based on android system |
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 |
Application publication date: 20161123 |
|
| RJ01 | Rejection of invention patent application after publication |