CN106951916A - A kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm - Google Patents
A kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm Download PDFInfo
- Publication number
- CN106951916A CN106951916A CN201710105718.0A CN201710105718A CN106951916A CN 106951916 A CN106951916 A CN 106951916A CN 201710105718 A CN201710105718 A CN 201710105718A CN 106951916 A CN106951916 A CN 106951916A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- potato
- image
- connection weight
- multiresolution
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
Abstract
The present invention relates to potato hierarchical algorithmses technical field, more particularly to a kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm, including:Step S101:If obtaining dried potato image;Step S102:Calculate connection weight of the potato image in different abstraction hierarchies between input layer and hidden layer respectively by multiresolution algorithm;Step S103:By connection weight described in Adaboots algorithm filterings, best features are obtained;Step S104:The result of image image classification is determined according to connection weight and best features.Using the present embodiment technical scheme, connection weight is calculated by multiresolution algorithm, time-consuming training and classifying step is effectively reduced.The outer shape of potato is classified by Adaboots algorithms again, finally has higher efficiency and faster speed to Potato Quality classification.
Description
Technical field
The present invention relates to potato hierarchical algorithmses technical field, it is more particularly to a kind of based on multiresolution algorithm and
The Potato Quality stage division of Adaboost algorithm.
Background technology
In Potato Quality classification, shape is a very important index, and national standard has strict regulation.It is existing
There is the method that the different classifications for potato external sort are proposed in technology, but all occur inefficiency in various degree
The slower defect with hierarchical speed.
The content of the invention
Embodiment of the present invention goal of the invention is to provide a kind of horse based on multiresolution algorithm and Adaboost algorithm
Bell potato quality grading method, can be using multiresolution algorithm and Adaboots algorithms to potato using the technical scheme
Outer shape is classified, and finally has higher efficiency and faster speed to Potato Quality classification.
In order to realize foregoing invention purpose, complete skill scheme of the invention is as follows:
A kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm, including following step
Suddenly:
If obtaining dried potato image;
Calculated respectively by multiresolution algorithm potato image in different abstraction hierarchies input layer and hidden layer it
Between connection weight;
By connection weight described in Adaboots algorithm filterings, best features are obtained;
The result of image image classification is determined according to connection weight and best features.
It is preferred that, the step:Potato image is calculated respectively by multiresolution algorithm in different abstraction hierarchies
Connection weight between middle input layer and hidden layer, specifically comprises the following steps
It is to approach part and detail section by potato picture breakdown;
In new abstraction hierarchy, will obtain described in approach decomposed and approach part and new detail section to be new;
Detail section and new detail section are defined as to the connection weight between input layer and hidden layer.
It is preferred that, the step:By connection weight described in Adaboots algorithm filterings, best features are obtained;Specifically
Comprise the following steps
If the connection weight is m group training datas, (x1,y1) ... ... (xm, ym), wherein
xi(instance) ∈ X, yi(classification) ∈ Y={ -1 ,+1 }
Initialize D1(i)=1/m, D1(i) weight for being training sample i;
If best features quantity is t, as t=1 ..., T, grader hi:X → [- 1 1], h is grader;
If ej<0.5 continues following steps, otherwise stops;
If t-th of threshold values is βt, select βt∈R:εtFor grader htWeighting fault rate;
Next grader:
Wherein ZtFor normalization factor, D is madet+1As distribution function;
Obtain last grader:
The best features of each class image are obtained, the Weak Classifier that f, threshold value beta and polarity p are constituted is characterized:
It is preferred that, the step:The result of image classification is determined according to connection weight and best features, it is specific include with
Lower step
If the result of image classification is H (Y):
When H (Y) is equal to 1, it is determined that current potato image meets current class, otherwise it is defined as not meeting.
Therefore, using the present embodiment technical scheme, connection weight is calculated by multiresolution algorithm, effectively subtracted
The training taken and classifying step are lacked.The outer shape of potato is classified by Adaboots algorithms again, it is final right
Potato Quality classification has higher efficiency and faster speed.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, without having to pay creative labor, may be used also
To obtain other accompanying drawings according to these accompanying drawings.
Fig. 1 is the program flow diagram that the embodiment of the present invention 1 is provided.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.It is based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made
Embodiment, belongs to the scope of protection of the invention.
Embodiment 1:
As shown in figure 1, the present embodiment provides a kind of potato product based on multiresolution algorithm and Adaboost algorithm
Matter stage division, comprises the following steps:
Step S101:If obtaining dried potato image;
Step S102:Calculate potato image input layer in different abstraction hierarchies respectively by multiresolution algorithm
Connection weight between hidden layer;
Step S103:By connection weight described in Adaboots algorithm filterings, best features are obtained;
Step S104:The result of image image classification is determined according to connection weight and best features.
It is more specifically as follows as preferred technical scheme:
Between step S102 and S103, the present embodiment can using value [0 1] sigmoid functions as swash
Function living.
In step s 102:Specifically comprise the following steps
It is to approach part and detail section by potato picture breakdown;
In new abstraction hierarchy, will obtain described in approach decomposed and approach part and new detail section to be new;
Detail section and new detail section are defined as to the connection weight between input layer and hidden layer.
Above step uses the fast wavelet transform (FWT) in multiresolution analysis, effectively reduces time-consuming training
And classifying step.Detail section therein includes level detail, vertical detail and diagonal details.
In step s 103:Specifically comprise the following steps
By connection weight described in Adaboots algorithm filterings, best features are obtained;Specifically comprise the following steps
If the connection weight is m group training datas, (x1,y1) ... ... (xm, ym), wherein
xi(instance) ∈ X, yi(classification) ∈ Y={ -1 ,+1 }
Initialize D1(i)=1/m, D1(i) weight for being training sample i;Which dictates that the sample is selected as compositional classification
The probability of device.
If best features quantity is t, as t=1 ..., T, grader hi:X → [- 1 1], h is grader;This will most
Reduce to limits distribution DtThe error brought:
If ej<0.5 continues following steps, otherwise stops;
If t-th of threshold values is βt, select βt∈R:εtFor grader htWeighting fault rate;
Next grader:
Wherein ZtFor normalization factor, D is madet+1As distribution function;
Obtain last grader:
The best features of each class image are obtained, the Weak Classifier that f, threshold value beta and polarity p are constituted is characterized:
Threshold value beta will be used as image of the connection weight between hidden layer and output layer for each class of determination.
In step S104:Image Y as classification is classified, it is necessary to set up a neutral net, wherein should
The hidden neuron connection weight of network is by that can represent that all features of all categories image are constituted, specifically including following
Step
If the result of image classification is H (Y):
When H (Y) is equal to 1, it is determined that current potato image meets current class, otherwise it is defined as not meeting.
The embodiment of the present invention using the fast wavelet transform based on multiresolution algorithm different levels it is abstract in carry
The feature of potato image is taken out, secondly, we select best feature and with this to corresponding using Adaboost algorithm
Potato image is classified.From the point of view of experimental result, classifying quality is very good, and demonstrate proposed based on many points
Resolution is analyzed and the simple deep learning neural network structure of Adaboost algorithm is efficient.
Embodiments described above, does not constitute the restriction to the technical scheme protection domain.It is any in above-mentioned implementation
Modification, equivalent and improvement made within the spirit and principle of mode etc., should be included in the protection model of the technical scheme
Within enclosing.
Claims (4)
1. a kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm, it is characterised in that bag
Include following steps:
If obtaining dried potato image;
Potato image is calculated respectively by multiresolution algorithm in different abstraction hierarchies between input layer and hidden layer
Connection weight;
By connection weight described in Adaboots algorithm filterings, best features are obtained;
The result of image image classification is determined according to connection weight and best features.
2. a kind of Potato Quality classification based on multiresolution algorithm and Adaboost algorithm according to claim 1
Method, it is characterised in that:
The step:Calculated respectively by multiresolution algorithm potato image in different abstraction hierarchies input layer with it is hidden
Connection weight between layer, specifically comprises the following steps
It is to approach part and detail section by potato picture breakdown;
In new abstraction hierarchy, will obtain described in approach decomposed and approach part and new detail section to be new;
Detail section and new detail section are defined as to the connection weight between input layer and hidden layer.
3. a kind of Potato Quality classification based on multiresolution algorithm and Adaboost algorithm according to claim 2
Method, it is characterised in that:
The step:By connection weight described in Adaboots algorithm filterings, best features are obtained;Specifically comprise the following steps
If the connection weight is m group training datas, (x1,y1) ... ... (xm, ym), wherein
xi(instance) ∈ X, yi(classification) ∈ Y={ -1 ,+1 }
Initialize D1(i)=1/m, D1(i) weight for being training sample i;
If best features quantity is t, as t=1 ..., T, grader hi:X → [- 1 1], h is grader;
If ej<0.5 continues following steps, otherwise stops;
If t-th of threshold values is βt, select βt∈R:εtFor grader htWeighting fault rate;
Next grader:
Wherein ZtFor normalization factor, D is madet+1As distribution function;
Obtain last grader:
The best features of each class image are obtained, the Weak Classifier that f, threshold value beta and polarity p are constituted is characterized:
4. a kind of Potato Quality classification based on multiresolution algorithm and Adaboost algorithm according to claim 3
Method, it is characterised in that:
The step:The result of image classification is determined according to connection weight and best features, is specifically comprised the following steps
If the result of image classification is H (Y):
When H (Y) is equal to 1, it is determined that current potato image meets current class, otherwise it is defined as not meeting.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2016107976790 | 2016-08-31 | ||
CN201610797679 | 2016-08-31 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106951916A true CN106951916A (en) | 2017-07-14 |
Family
ID=59467716
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710105718.0A Withdrawn CN106951916A (en) | 2016-08-31 | 2017-02-26 | A kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106951916A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112764110A (en) * | 2020-07-09 | 2021-05-07 | 五季数据科技(北京)有限公司 | Clustered seismic facies analysis method based on limiting Boltzmann machine feature coding |
CN115841600A (en) * | 2023-02-23 | 2023-03-24 | 山东金诺种业有限公司 | Deep learning-based sweet potato appearance quality classification method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101344928A (en) * | 2007-07-12 | 2009-01-14 | 佳能株式会社 | Method and apparatus for confirming image area and classifying image |
CN104056790A (en) * | 2013-03-19 | 2014-09-24 | 青岛农业大学 | Intelligent potato sorting method and apparatus |
CN104597052A (en) * | 2015-02-09 | 2015-05-06 | 淮阴工学院 | High-speed lossless potato grading detection method and system based on multi-characteristic fusion |
CN105678755A (en) * | 2015-12-31 | 2016-06-15 | 青岛歌尔声学科技有限公司 | Product state detection method and system based on Adaboost algorithm |
-
2017
- 2017-02-26 CN CN201710105718.0A patent/CN106951916A/en not_active Withdrawn
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101344928A (en) * | 2007-07-12 | 2009-01-14 | 佳能株式会社 | Method and apparatus for confirming image area and classifying image |
CN104056790A (en) * | 2013-03-19 | 2014-09-24 | 青岛农业大学 | Intelligent potato sorting method and apparatus |
CN104597052A (en) * | 2015-02-09 | 2015-05-06 | 淮阴工学院 | High-speed lossless potato grading detection method and system based on multi-characteristic fusion |
CN105678755A (en) * | 2015-12-31 | 2016-06-15 | 青岛歌尔声学科技有限公司 | Product state detection method and system based on Adaboost algorithm |
Non-Patent Citations (1)
Title |
---|
ASMA ELADEL等: "Deep learning with shallow architecture for image classification", 《2015 INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING & SIMULATION(HPCS)》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112764110A (en) * | 2020-07-09 | 2021-05-07 | 五季数据科技(北京)有限公司 | Clustered seismic facies analysis method based on limiting Boltzmann machine feature coding |
CN115841600A (en) * | 2023-02-23 | 2023-03-24 | 山东金诺种业有限公司 | Deep learning-based sweet potato appearance quality classification method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108182454B (en) | Security check identification system and control method thereof | |
Lempitsky et al. | Pylon model for semantic segmentation | |
CN106845510A (en) | Chinese tradition visual culture Symbol Recognition based on depth level Fusion Features | |
CN107563428A (en) | Classification of Polarimetric SAR Image method based on generation confrontation network | |
CN107330446A (en) | A kind of optimization method of depth convolutional neural networks towards image classification | |
CN106372648A (en) | Multi-feature-fusion-convolutional-neural-network-based plankton image classification method | |
CN103020122A (en) | Transfer learning method based on semi-supervised clustering | |
CN105512661A (en) | Multi-mode-characteristic-fusion-based remote-sensing image classification method | |
CN107301400A (en) | A kind of semantic semi-supervised video picture segmentation method being oriented to | |
CN102855633A (en) | Anti-noise quick fuzzy-clustering digital image segmentation method | |
CN102013017A (en) | Method for roughly sorting high-resolution remote sensing image scene | |
CN104850890A (en) | Method for adjusting parameter of convolution neural network based on example learning and Sadowsky distribution | |
CN111145145B (en) | Image surface defect detection method based on MobileNet | |
CN109002463A (en) | A kind of Method for text detection based on depth measure model | |
CN102867183A (en) | Method and device for detecting littered objects of vehicle and intelligent traffic monitoring system | |
CN113468291B (en) | Patent automatic classification method based on patent network representation learning | |
CN108564166A (en) | Based on the semi-supervised feature learning method of the convolutional neural networks with symmetrical parallel link | |
CN110751195A (en) | Fine-grained image classification method based on improved YOLOv3 | |
CN110084284A (en) | Target detection and secondary classification algorithm and device based on region convolutional neural networks | |
CN110264407A (en) | Image Super-resolution model training and method for reconstructing, device, equipment and storage medium | |
Zhu et al. | Learning reconfigurable scene representation by tangram model | |
CN110009005A (en) | A kind of net flow assorted method based on feature strong correlation | |
CN111104975B (en) | Credit evaluation method based on breadth learning | |
CN106951916A (en) | A kind of Potato Quality stage division based on multiresolution algorithm and Adaboost algorithm | |
CN108090913A (en) | A kind of image, semantic dividing method based on object level Gauss-Markov random fields |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170714 |
|
WW01 | Invention patent application withdrawn after publication |