CN107657633A - A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition - Google Patents

A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition Download PDF

Info

Publication number
CN107657633A
CN107657633A CN201710900797.4A CN201710900797A CN107657633A CN 107657633 A CN107657633 A CN 107657633A CN 201710900797 A CN201710900797 A CN 201710900797A CN 107657633 A CN107657633 A CN 107657633A
Authority
CN
China
Prior art keywords
mrow
soil
neural network
msub
image
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
Application number
CN201710900797.4A
Other languages
Chinese (zh)
Inventor
王娜
吴健宇
吴芝路
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.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of 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 Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN201710900797.4A priority Critical patent/CN107657633A/en
Publication of CN107657633A publication Critical patent/CN107657633A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G06T7/45—Analysis of texture based on statistical description of texture using co-occurrence matrix computation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/28—Measuring arrangements characterised by the use of optical techniques for measuring areas
    • G01B11/285—Measuring arrangements characterised by the use of optical techniques for measuring areas using photoelectric detection means
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • 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
    • G06T2207/30181—Earth observation
    • G06T2207/30188—Vegetation; Agriculture

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition, the present invention is to solve because artificial drawstring method measurement efficiency is low, error is big, the shortcomings of labor intensity is big, and proposed during measuring and calculating due to the limitation that human factor is brought.Including：The post-job view data of straw-returning is gathered every the set time；The image block of default size is intercepted in the view data of acquisition as sample；Calculate five kinds of texture eigenvalues in the sample：Energy, the moment of inertia, entropy, correlation, inverse difference moment；BP neural network of the structure with input layer, hidden layer and output layer；Neutral net is trained：Utilize Processing with Neural Network testing image：And binary conversion treatment is carried out to image according to classification results；Improving straw mulching rate is calculated according to image total pixel number amount ratio is accounted for according to white pixel point quantity.The present invention is applied to soil improving straw mulching rate and calculated.

Description

A kind of soil improving straw mulching rate based on BP neural network and sensor data acquisition is surveyed Calculation method
Technical field
The present invention relates to the soil improving straw mulching rate calculating field based on BP neural network, and in particular to one kind is based on BP god Soil improving straw mulching rate computational methods through network and sensor data acquisition.
Background technology
Straw-returning is should not directly make the stalk (wheat straw, corn stalk and rice straw etc.) of feed directly or accumulation is rotten A kind of method in being manured into soil after ripe.Straw-returning is that the volume increase of a culture fertility of most attention in the world today is arranged Apply, also have getting fat production-increasing function while the atmosphere pollution caused by having prevented crop straw burning, be important agricultural practice.Mesh The main calculation methods of preceding soil improving straw mulching rate are the methods using manual measurement, and this method is referred to as " drawstring method ".By Low in this method efficiency, error is big, and labor intensity is big, so the improving straw mulching rate often measured is inaccurate.The present invention The image processing techniques in modern information technologies is effectively utilized, will all be entered per a image information by the method for BP neural network Row pixel travels through, so as to distinguish soil area and stalk area.This computational methods efficiency high, error is small, is a kind of more preferable Solution.
The content of the invention
It is an object of the invention to overcome because artificial drawstring method measurement efficiency is low, the shortcomings of error is big, and labor intensity is big, And due to limitation that human factor is brought during measuring and calculating, there is provided one kind is adopted based on BP neural network and sensing data The improving straw mulching rate Area computing method of collection.
In order to solve the problems existing in background technology, the present invention uses solution below：
A kind of straw-returning coverage rate measuring method based on BP neural network and sensor data acquisition, it includes following Step：
Step 1：The post-job view data of straw-returning is gathered every the set time；
Step 2：The image block of default size is intercepted in the view data of acquisition as sample；
Step 3：Calculate five kinds of texture eigenvalues in the sample：Energy, the moment of inertia, entropy, correlation, inverse difference moment；
Step 4：BP neural network of the structure with input layer, hidden layer and output layer；
Step 5：Neutral net is trained：During training, transmission function selection s type nonlinear functions；Neutral net Training result, we set output layer soil value and are normalized to 0.1, and output layer stalk value is normalized to 0.9；If The respectively desired output of soil and stalk, i.e.,O1k, O2kThe respectively reality of soil and stalk Network exports；
Have for soil：
Have for stalk：
And
When above three formula is met simultaneously, neutral net deconditioning；Otherwise, neutral net back transfer, Weights are changed, until meeting above formula, now training terminates；
Step 6：Utilize Processing with Neural Network testing image：For each pixel in shooting image, to around it 3 × 3 region carries out feature extraction, utilizes neural computing output valve；If output valve is more than 0.1-0.05 and is less than During 0.1+0.05, pixel value is set to 0；If output valve is more than 0.9-0.05 and is less than 0.9+0.05, pixel value is set to 255；
Step 7：Calculate improving straw mulching rate：Image after statistical disposition shares m white pixel point, according to formulaCalculate improving straw mulching rate；Wherein N is shooting image pixel total quantity.
Preferably, step 2 is specially：The sample of 40 pure soil and 40 pure stalks is intercepted in the view data of acquisition This, sample is the image block of 30 × 30 sizes.
Preferably, step 3 is specially：For 30 × 30 image block, the image is horizontal, longitudinal direction is respectively X, Y-axis, quantifies Series L=256, gray value i, j interval are (0, L-1)；The point (x, y) that gray value is i in image leaves original along θ directions Carry out position δ (Dx,Dy) after, new position point (x+Dx,y+Dy) on gray value be j probability be pθ(i, j), its expression formula are：
Gray level co-occurrence matrixes are relevant with direction, define θ 4 different directions, i.e.,：
0°(Dx=d, Dy=0), 45 ° of (Dx=d, Dy=d), 90 ° of (Dx=0, Dy=d), 135 ° of (Dx=-d, Dy=d)
Wherein step-length d is minimum pixel spacing；Following five texture eigenvalues of definition：
Energy：
The moment of inertia：
Entropy：
Correlation：
In formula,
Inverse difference moment：
The texture eigenvalue for extracting 40 soil samples and 40 stalk samples is calculated by above formula.
Preferably, step 4 is specially：BP neural network of the structure with input layer, hidden layer and output layer；Wherein, Input layer shares five neurons, the input of five kinds of features for realizing a sample；Hidden layer shares four neurons；It is defeated Go out layer and share two neurons：Soil and stalk are represented respectively.
Preferably, step 5 is specially：
Neutral net is trained：During training, input value：x1k=ASM, x2k=CON, x3k=ENT, x4k=4COR, x5k =M (d, θ), xkFor k-th of learning objective of input layer；In the neutral net, transmission function selection s type nonlinear functions； Setting output layer soil value is normalized to 0.1, and output layer stalk value is normalized to 0.9；IfRespectively soil with The desired output of stalk, i.e.,And O1k, O2kThe respectively real network output of soil and stalk；
Have for soil：
Have for stalk：
And
When above three formula is met simultaneously, neutral net deconditioning；Otherwise, neutral net back transfer, Weights are changed, until meeting above formula, now training terminates.
Preferably, step 6 is specially：
Utilize Processing with Neural Network testing image：320 × 240 image of shooting is carried out with the neutral net trained Processing, to each pixel, feature extraction is carried out to around it 3 × 3 region, utilizes neural computing output valve；Such as When fruit output valve is more than 0.1-0.05 and is less than 0.1+0.05, pixel value is set to 0；If output valve be more than 0.9-0.05 and During less than 0.9+0.05, pixel value is set to 255.
The beneficial effects of the invention are as follows：1st, the present invention does not need manual measurement, more efficient, and error is smaller, labor intensity It is smaller；2nd, the present invention carries out automatic detection by image procossing, eliminates the limitation that human factor is brought；3rd, the present invention improves The degree of accuracy of improving straw mulching rate calculated value.
Brief description of the drawings
The present invention is further described with example below in conjunction with the accompanying drawings.
Fig. 1 is the straw-returning coverage rate measuring method stream based on BP neural network and sensor data acquisition of the present invention Cheng Tu.
Fig. 2 is sample collection schematic diagram.
Fig. 3 is single neuronal structure figure.
Fig. 4 is neural network structure figure.
Embodiment
The invention will be further described with reference to the accompanying drawings and detailed description：
Fig. 1 is the straw-returning coverage rate measuring method stream based on BP neural network and sensor data acquisition of the present invention Cheng Tu.It comprises the following steps：
Step 1：Data acquisition：Gathered by the first-class sensor of shooting on agricultural equipment every the set time The post-job view data of straw-returning.
Step 2：Sample is chosen：The sample of 40 pure soil and 40 pure stalks is intercepted in the image of acquisition, it is desirable to sample This is the image block that size is 30 × 30
Step 3：Sample characteristics extract：For 30 × 30 sample image, the image is horizontal, and longitudinal direction is respectively X, Y-axis, is measured Change series L=256, gray value i, j interval are (0, L-1)；The point (x, y) that gray value is i in image leaves along θ directions Origin-location δ (Dx,Dy) after, new position point (x+Dx,y+Dy) on gray value be j probability be pθ(i, j), its expression formula are：
Gray level co-occurrence matrixes are relevant with direction, θ 4 different directions defined in research, i.e.,：0°(Dx=d, Dy=0), 45°(Dx=d, Dy=d), 90 ° of (Dx=0, Dy=d), 135 ° of (Dx=-d, Dy=d)
Reflect the information of whole pictures comprehensively by this 4 directions.Step-length d takes minimum pixel spacing defined in experiment, Its value is 1, can so ensure more fully to read pictorial information.And the characteristic quantity of texture is carried on the basis of gray level co-occurrence matrixes Go out, can intuitively reflect the second degree statistics of pictorial information, and the selected amount of BP neural network input layer, present invention selection Following five texture eigenvalues：
Energy (angular second moment)：
The moment of inertia：
Entropy：
Correlation：
In formula,
Inverse difference moment：
The texture eigenvalue for extracting 40 soil samples and 40 stalk samples is calculated by above formula.
Step 4：The structure of neutral net：The BP neural network that the present invention is built is divided into three big layers：Input layer, imply Layer and output layer.Wherein, input layer shares five neurons, realizes the input of five kinds of features of a sample；Hidden layer exists There are four neurons in this secondary design；Output layer shares two neurons：Soil and stalk are represented respectively.
Step 5：The training of neutral net：During training, input value：x1k=ASM, x2k=CON, x3k=ENT, x4k=4COR, x5k=M (d, θ), xkFor k-th of learning objective of input layer；In the neutral net, transmission function we select s types non-linear Function；The training result of neutral net, we set output layer soil value and are normalized to 0.1, and output layer stalk value normalizes For 0.9；IfThe respectively desired output of soil and stalk, i.e., And O1k, O2kRespectively Exported for the real network of soil and stalk；
Have for soil：
Have for stalk：
And have：
When above three formula is met simultaneously, neutral net deconditioning；Otherwise, neutral net back transfer, Weights are changed, until meeting above formula, now training terminates.
Step 6：Utilize Processing with Neural Network testing image：By 320 × 240 image of the shooting nerve trained Network is handled, and to each pixel, feature extraction is carried out to around it 3 × 3 region, defeated using neural computing Go out value.If output valve meets soil characteristic, when it is less than 0.5, black, pixel value 0 are entered as；If output valve meets Stalk characteristic, when it is more than 0.5, it is entered as white, pixel value 255；So it is achieved that image binaryzation.
Step 7：Calculate improving straw mulching rate：Image after statistical disposition shares m white pixel point, according to formulaCalculate improving straw mulching rate.
It is sample collection schematic diagram such as Fig. 2.Be divided into figure stalk area and soil area (be only schematic diagram, soil in real image Earth and stalk can not be distinguished easily), stalk area and soil area can multiple repairing weld, it is necessary to which the sample for ensureing to gather does not weigh Multiple, size is 30 × 30, collects 40 soil samples and 40 stalk samples.
It is single neuronal structure figure such as Fig. 3, inputs as n characteristic value, n weighted input is summed, then pass through transmission Function, you can obtain the output valve of single neuron.In the present invention, transmission function have selected s type nonlinear functions,
It is neural network structure figure such as Fig. 4.The BP neural network that the present invention is built is divided into three big layers：Input layer, imply Layer and output layer.Wherein, input layer shares five neurons, realizes the input of five kinds of features of a sample, i.e. x1k= ASM, x2k=CON, x3k=ENT, x4k=4COR, x5k=M (d, θ), xkFor k-th of learning objective of input layer；Hidden layer is current There are four neurons in design；Output layer shares two neurons：Soil and stalk are represented respectively.
Present invention contrast prior art, has the advantages that：Traditional soil improving straw mulching rate computational methods are due to people Work factor, sampling point distributions are uneven, cause coverage rate computational accuracy relatively low.It is big for hand labor intensity, calculate inaccurate The shortcomings of, the present invention is handled the image information of upload using image procossing and BP neural network thought, so can be big Reduce unnecessary manual labor greatly, ensure that the degree of accuracy of improving straw mulching rate calculated value.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.

Claims (6)

  1. A kind of 1. soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition, it is characterised in that Comprise the following steps：
    Step 1：The post-job view data of straw-returning is gathered every the set time；
    Step 2：The image block of default size is intercepted in the view data of acquisition as sample；
    Step 3：Calculate five kinds of texture eigenvalues in the sample：Energy, the moment of inertia, entropy, correlation, inverse difference moment；
    Step 4：BP neural network of the structure with input layer, hidden layer and output layer；
    Step 5：Neutral net is trained：During training, transmission function selection s type nonlinear functions；The training of neutral net As a result, we set output layer soil value and are normalized to 0.1, and output layer stalk value is normalized to 0.9；IfRespectively For soil and the desired output of stalk, i.e.,O1k, O2kThe respectively real network of soil and stalk Output；
    Have for soil：
    Have for stalk：
    And
    When above three formula is met simultaneously, neutral net deconditioning；Otherwise, neutral net back transfer, modification Weights, until meeting above formula, now training terminates；
    Step 6：Utilize Processing with Neural Network testing image：For each pixel in shooting image, to around it 3 × 3 Region carry out feature extraction, utilize neural computing output valve；If output valve is more than 0.1-0.05 and is less than 0.1+ When 0.05, pixel value is set to 0；If output valve is more than 0.9-0.05 and is less than 0.9+0.05, pixel value is set to 255；
    Step 7：Calculate improving straw mulching rate：Image after statistical disposition shares m white pixel point, according to formula Calculate improving straw mulching rate；Wherein N is shooting image pixel total quantity.
  2. 2. the soil improving straw mulching rate measuring and calculating side according to claim 1 based on BP neural network and sensor data acquisition Method, it is characterised in that step 2 is specially：
    The sample of 40 pure soil and 40 pure stalks is intercepted in the view data of acquisition, sample is the image of 30 × 30 sizes Block.
  3. 3. the soil improving straw mulching rate measuring and calculating side according to claim 2 based on BP neural network and sensor data acquisition Method, it is characterised in that step 3 is specially：
    For 30 × 30 image block, the image is horizontal, longitudinal direction is respectively X, Y-axis, quantifies series L=256, gray value i, j value Section is (0, L-1)；The point (x, y) that gray value is i in image leaves origin-location δ (D along θ directionsx,Dy) after, new position point (x+Dx,y+Dy) on gray value be j probability be pθ(i, j), its expression formula are：
    Gray level co-occurrence matrixes are relevant with direction, define θ 4 different directions, i.e.,：
    0°(Dx=d, Dy=0), 45 ° of (Dx=d, Dy=d), 90 ° of (Dx=0, Dy=d), 135 ° of (Dx=-d, Dy=d)
    Wherein step-length d is minimum pixel spacing；Following five texture eigenvalues of definition：
    Energy：
    The moment of inertia：
    Entropy：
    Correlation：
    In formula,
    <mrow> <msub> <mi>&amp;mu;</mi> <mn>2</mn> </msub> <mo>=</mo> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msubsup> <mi>j&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msub> <mi>p</mi> <mi>&amp;theta;</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow>
    <mrow> <msub> <mi>&amp;sigma;</mi> <mn>1</mn> </msub> <mo>=</mo> <msqrt> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msup> <mrow> <mo>(</mo> <mi>i</mi> <mo>-</mo> <msub> <mi>&amp;mu;</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msub> <mi>p</mi> <mi>&amp;theta;</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </msqrt> </mrow>
    <mrow> <msub> <mi>&amp;sigma;</mi> <mn>2</mn> </msub> <mo>=</mo> <msqrt> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msup> <mrow> <mo>(</mo> <mi>j</mi> <mo>-</mo> <msub> <mi>&amp;mu;</mi> <mn>2</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>L</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msub> <mi>p</mi> <mi>&amp;theta;</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </msqrt> </mrow>
    Inverse difference moment：
    The texture eigenvalue for extracting 40 soil samples and 40 stalk samples is calculated by above formula.
  4. 4. the soil improving straw mulching rate measuring and calculating side according to claim 3 based on BP neural network and sensor data acquisition Method, it is characterised in that step 4 is specially：BP neural network of the structure with input layer, hidden layer and output layer；Wherein, Input layer shares five neurons, the input of five kinds of features for realizing a sample；Hidden layer shares four neurons；It is defeated Go out layer and share two neurons：Soil and stalk are represented respectively.
  5. 5. the soil improving straw mulching rate calculating side according to claim 4 based on BP neural network and sensor data acquisition Method, it is characterised in that step 5 is specially：
    Neutral net is trained：During training, input value：x1k=ASM, x2k=CON, x3k=ENT, x4k=4COR, x5k=M (d, θ), xkFor k-th of learning objective of input layer；In the neutral net, transmission function selection s type nonlinear functions；Set defeated Go out a layer soil value and be normalized to 0.1, and output layer stalk value is normalized to 0.9；IfRespectively soil and stalk Desired output, i.e.,And O1k, O2kThe respectively real network output of soil and stalk；
    Have for soil：
    Have for stalk：
    And
    When above three formula is met simultaneously, neutral net deconditioning；Otherwise, neutral net back transfer, modification Weights, until meeting above formula, now training terminates.
  6. 6. the soil improving straw mulching rate measuring and calculating side according to claim 5 based on BP neural network and sensor data acquisition Method, it is characterised in that step 6 is specially：
    Utilize Processing with Neural Network testing image：By 320 × 240 image of shooting with the neutral net trained Reason, to each pixel, feature extraction is carried out to around it 3 × 3 region, utilizes neural computing output valve；If When output valve is more than 0.1-0.05 and is less than 0.1+0.05, pixel value is set to 0；If output valve is more than 0.9-0.05 and is less than During 0.9+0.05, pixel value is set to 255.
CN201710900797.4A 2017-09-28 2017-09-28 A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition Pending CN107657633A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710900797.4A CN107657633A (en) 2017-09-28 2017-09-28 A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710900797.4A CN107657633A (en) 2017-09-28 2017-09-28 A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition

Publications (1)

Publication Number Publication Date
CN107657633A true CN107657633A (en) 2018-02-02

Family

ID=61117097

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710900797.4A Pending CN107657633A (en) 2017-09-28 2017-09-28 A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition

Country Status (1)

Country Link
CN (1) CN107657633A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109166121A (en) * 2018-09-12 2019-01-08 西南大学 Fissured expansive soils textural characteristics quantification acquisition methods
CN109272252A (en) * 2018-10-17 2019-01-25 郑州轻工业学院 A kind of accounting method based on polymorphic straw resource utilization value
CN110509186A (en) * 2019-08-29 2019-11-29 华中科技大学 A kind of robot grinding and polishing quality characterization method based on processing vibration performance
CN110554352A (en) * 2019-09-11 2019-12-10 哈尔滨工业大学 Method for estimating direction of arrival of interference source of aerospace measurement and control system based on VGG16 neural network
CN117333494A (en) * 2023-12-01 2024-01-02 辽宁牧龙科技有限公司 Deep learning-based straw coverage rate detection method and system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102222267A (en) * 2010-04-14 2011-10-19 同济大学 Remote-sensing classification processing method for neural network by genetic algorithm
CN103529189A (en) * 2013-06-28 2014-01-22 四川农业大学 Soil organic matter space distribution predication method based on qualitative and quantitative auxiliary variables
CN105528523A (en) * 2015-12-18 2016-04-27 中国科学院遥感与数字地球研究所 Soil thickness inversion method based on remote sensing data
CN105675807A (en) * 2016-01-07 2016-06-15 中国农业大学 Evaluation method of atrazine residue based on BP neural network
CN105938563A (en) * 2016-04-14 2016-09-14 北京工业大学 Weld surface defect identification method based on image texture
CN106951836A (en) * 2017-03-05 2017-07-14 北京工业大学 Crop cover degree extracting method based on priori threshold optimization convolutional neural networks
CN107067043A (en) * 2017-05-25 2017-08-18 哈尔滨工业大学 A kind of diseases and pests of agronomic crop detection method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102222267A (en) * 2010-04-14 2011-10-19 同济大学 Remote-sensing classification processing method for neural network by genetic algorithm
CN103529189A (en) * 2013-06-28 2014-01-22 四川农业大学 Soil organic matter space distribution predication method based on qualitative and quantitative auxiliary variables
CN105528523A (en) * 2015-12-18 2016-04-27 中国科学院遥感与数字地球研究所 Soil thickness inversion method based on remote sensing data
CN105675807A (en) * 2016-01-07 2016-06-15 中国农业大学 Evaluation method of atrazine residue based on BP neural network
CN105938563A (en) * 2016-04-14 2016-09-14 北京工业大学 Weld surface defect identification method based on image texture
CN106951836A (en) * 2017-03-05 2017-07-14 北京工业大学 Crop cover degree extracting method based on priori threshold optimization convolutional neural networks
CN107067043A (en) * 2017-05-25 2017-08-18 哈尔滨工业大学 A kind of diseases and pests of agronomic crop detection method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
李洪文 等: "基于人工神经网络的田间秸秆覆盖率检测系统", 《农业机械学报》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109166121A (en) * 2018-09-12 2019-01-08 西南大学 Fissured expansive soils textural characteristics quantification acquisition methods
CN109272252A (en) * 2018-10-17 2019-01-25 郑州轻工业学院 A kind of accounting method based on polymorphic straw resource utilization value
CN110509186A (en) * 2019-08-29 2019-11-29 华中科技大学 A kind of robot grinding and polishing quality characterization method based on processing vibration performance
CN110554352A (en) * 2019-09-11 2019-12-10 哈尔滨工业大学 Method for estimating direction of arrival of interference source of aerospace measurement and control system based on VGG16 neural network
CN117333494A (en) * 2023-12-01 2024-01-02 辽宁牧龙科技有限公司 Deep learning-based straw coverage rate detection method and system
CN117333494B (en) * 2023-12-01 2024-02-27 辽宁牧龙科技有限公司 Deep learning-based straw coverage rate detection method and system

Similar Documents

Publication Publication Date Title
CN107657633A (en) A kind of soil improving straw mulching rate measuring method based on BP neural network and sensor data acquisition
CN104598908B (en) A kind of crops leaf diseases recognition methods
CN107576618B (en) Rice panicle blast detection method and system based on deep convolutional neural network
CN110287944B (en) Crop pest monitoring method based on multispectral remote sensing image of deep learning
CN105004320B (en) A kind of high score satellite data land table vegetation coverage inversion method and system
CN107292343A (en) A kind of Classification of hyperspectral remote sensing image method based on six layers of convolutional neural networks and spectral space information consolidation
CN109961024A (en) Wheat weeds in field detection method based on deep learning
CN107392130A (en) Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks
CN105825177A (en) Remote-sensing crop disease identification method based on time phase and spectrum information and habitat condition
CN113657326A (en) Weed detection method based on multi-scale fusion module and feature enhancement
Roumagnac et al. Spatial and temporal analyses of bacterial blight of onion caused by Xanthomonas axonopodis pv. allii
CN106600595A (en) Human body characteristic dimension automatic measuring method based on artificial intelligence algorithm
CN107909109A (en) SAR image sorting technique based on conspicuousness and multiple dimensioned depth network model
CN106845497A (en) Maize in Earlier Stage image damage caused by a drought recognition methods based on multi-feature fusion
CN106529458A (en) Deep neural network space spectrum classification method for high-spectral image
CN107358203A (en) A kind of High Resolution SAR image classification method based on depth convolution ladder network
CN114067207A (en) Vegetable seedling field weed detection method based on deep learning and image processing
CN105893977A (en) Rice mapping method based on self-adaptive feature selection
CN108710864A (en) Winter wheat Remotely sensed acquisition method based on various dimensions identification and image noise reduction processing
CN109670509A (en) Winter wheat seedling stage growing way parameter evaluation method and system based on convolutional neural networks
Zhang et al. Evaluation of efficacy of fungicides for control of wheat fusarium head blight based on digital imaging
CN109977901A (en) A kind of corn weeds in field recognition methods based on unmanned aerial vehicle remote sensing
Xiao et al. Citrus greening disease recognition algorithm based on classification network using TRL-GAN
CN114693616A (en) Rice disease detection method, equipment and medium based on improved target detection model and convolutional neural network
Parreiras et al. Using unmanned aerial vehicle and machine learning algorithm to monitor leaf nitrogen in coffee

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
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20180202

WD01 Invention patent application deemed withdrawn after publication