基于CNN神經(jīng)網(wǎng)絡(luò)的小麥不完善粒高光譜檢測
本文選題:小麥 + 不完善粒 ; 參考:《食品科學(xué)》2017年24期
【摘要】:利用高光譜成像技術(shù)對(duì)小麥不完善粒進(jìn)行無損檢測。以932個(gè)小麥為樣本,其中正常粒樣本486個(gè)、破損粒樣本170個(gè)、蟲蝕粒樣本149個(gè)及黑胚粒樣本127個(gè)為研究對(duì)象,通過高光譜圖像采集系統(tǒng)采集樣本的光譜信息,然后從每個(gè)樣本的116個(gè)波段中選取30個(gè)波段,建立基于深度學(xué)習(xí)的卷積神經(jīng)網(wǎng)絡(luò)(convolutional neural networks,CNN)模型。實(shí)驗(yàn)中的CNN采用2個(gè)卷積層,第1層采用大小為3×3的32個(gè)卷積核,第2層采用大小為5×5的64個(gè)卷積核,池化層采用最大池,激活函數(shù)采用修正線性單元,為避免過擬合,在全連接層后面接入dropout層,參數(shù)設(shè)置為0.5,其他卷積參數(shù)均為默認(rèn)值,得到校正集總識(shí)別率為100.00%,測試集總識(shí)別率為99.98%。最后,以支持向量機(jī)(support vector machine,SVM)為基線模型進(jìn)行對(duì)比,從116個(gè)波段中選取90個(gè)波段進(jìn)行建模,測試集總識(shí)別率為94.73%。通過實(shí)驗(yàn)對(duì)比可以看出,CNN模型比SVM模型識(shí)別率高。研究表明CNN模型能夠?qū)崿F(xiàn)對(duì)小麥不完善粒的準(zhǔn)確、快速、無損檢測。
[Abstract]:The nondestructive testing of wheat imperfect grains was carried out by using hyperspectral imaging technique. The spectral information of 932 wheat samples was collected by hyperspectral image acquisition system, including 486 normal grain samples, 170 damaged grain samples, 149 wormwood samples and 127 black embryo samples. Then, 30 bands are selected from 116 bands of each sample, and a convolution neural network (convolutional neural) model based on deep learning is established. In the experiment, CNN uses 2 convolution layers, the first layer uses 32 convolution cores with size 3 脳 3, the second layer uses 64 convolution cores with size 5 脳 5, the pool layer adopts the largest pool, and the activation function adopts modified linear unit to avoid overfitting. When the dropout layer is connected behind the full connection layer, the parameters are set to 0.5, the other convolution parameters are all the default values, the total recognition rate of the correction set is 100.00000 and the total recognition rate of the test set is 99.98. Finally, the support vector machine (support vector machine) is used as the baseline model, 90 bands are selected from 116 bands to model, and the total recognition rate of the test set is 94.73. Through the comparison of experiments, we can see that CNN model has higher recognition rate than SVM model. The results show that CNN model can be used for accurate, fast and nondestructive detection of wheat imperfect grains.
【作者單位】: 北京工商大學(xué)計(jì)算機(jī)與信息工程學(xué)院食品安全大數(shù)據(jù)技術(shù)北京市重點(diǎn)實(shí)驗(yàn)室;
【基金】:土壤植物機(jī)器系統(tǒng)技術(shù)國家重點(diǎn)實(shí)驗(yàn)室開放課題(2014-SKL-05) 北京工商大學(xué)兩科基金培育項(xiàng)目(LKJJ2015-22)
【分類號(hào)】:TP183;TP391.41
【相似文獻(xiàn)】
相關(guān)期刊論文 前5條
1 張玉榮;陳賽賽;周顯青;王偉宇;吳瓊;王海榮;;基于圖像處理和神經(jīng)網(wǎng)絡(luò)的小麥不完善粒識(shí)別方法研究[J];糧油食品科技;2014年03期
2 雷國偉,呂迎陽,紀(jì)安妮,吳孫桃,郭東輝;圖像特征的CNN提取方法及其應(yīng)用[J];計(jì)算機(jī)工程與應(yīng)用;2004年14期
3 楊懷恒;閔樂泉;;設(shè)計(jì)局部最大灰度值探測CNN模板的定理與應(yīng)用[J];計(jì)算機(jī)工程與應(yīng)用;2006年19期
4 陳瑞森;;數(shù)字CNN微處理器的指令集設(shè)計(jì)[J];現(xiàn)代電子技術(shù);2009年24期
5 沙莎;劉金珠;閔樂泉;;復(fù)合4鄰域圈提取CNN的魯棒性設(shè)計(jì)[J];計(jì)算機(jī)工程與應(yīng)用;2011年02期
相關(guān)會(huì)議論文 前1條
1 劉國華;張穎;陳子軍;陳子陽;;改進(jìn)的CNN搜索算法[A];第二十屆全國數(shù)據(jù)庫學(xué)術(shù)會(huì)議論文集(技術(shù)報(bào)告篇)[C];2003年
相關(guān)重要報(bào)紙文章 前1條
1 本報(bào)記者 馬佳;調(diào)查CNN“中國黑客”報(bào)道[N];北京科技報(bào);2008年
相關(guān)碩士學(xué)位論文 前1條
1 楊振杰;基于CNN的交通標(biāo)志識(shí)別方法研究[D];天津工業(yè)大學(xué);2017年
,本文編號(hào):2078046
本文鏈接:http://sikaile.net/kejilunwen/zidonghuakongzhilunwen/2078046.html