共查询到19条相似文献,搜索用时 78 毫秒
1.
2.
3.
4.
种子纯度是种子质量评价的重要指标,本研究针对高光谱图像检测效率难以满足大批量种子检测的问题,采用包络线去除-卷积神经网络(CR-CNN),开发一种快速、可靠的小麦种子分类方法。首先采用包络线去除法挑选出特征波段,然后结合Ghost模块、MobileNetV2模块压缩架构以及经典架构分别建立卷积神经网络分类模型,最后比较全波段和特征波段的模型检测结果。研究表明,使用包络线去除法之后,检测时间为原先的9.50%~12.87%,百个样本检测时间最快仅需要0.019 s,同时分类精度最高能达到96.125%。CR-CNN方法能够充分利用高光谱图像中的有效信息,快速且准确的鉴别小麦种子品种,为开发高精度小麦种子在线检测多光谱设备提供参考。 相似文献
5.
针对当下许多织物尚未有公定的统一名称、译名,尝试从不同视角对其进行命名和分类,以得到新的诠释。介绍了常见十五类机织物的基本品种,代表织物命名和分类方式,重在织物特性而非依照原料分类。 相似文献
6.
7.
8.
小麦中的酶类是影响面粉质量的重要因素。介绍了小麦中的酶类其消长情况,以及降低小麦中酶的负面影响的几种方式,最后,在回顾和展望国内外小麦品种研究和对优质小麦品种开发现状的基础上,阐述了促进我国优质小麦品种研究和开发的几点看法。 相似文献
9.
展望新世纪的优质小麦品种研究与开发(二) 总被引:20,自引:0,他引:20
小麦淀粉是决定小麦品质的固定之一,对小玫淀粉及其与食品品质,特别是东方面条品质的关系的研究进展和最新观点作了介绍 相似文献
10.
11.
Integrating varietal resistance with temperature manipulation during storage may provide a better option for protection of stored grains and may decrease reliance on the use of synthetic chemicals. The current study was conducted to determine the susceptibility of different varieties of wheat seed to the infestation by the granary weevil, Sitophilus granarius (L.), and rice weevil, Sitophilus oryzae (L.), at optimal (30 °C) and sub-optimal (19 °C) temperatures. Kernels of six wheat varieties namely, Danda'a, Digalu, ET-13-A2, Kakaba, Millennium, and Pavon-76 were examined over a period of 90 d. Significant interactions were detected between wheat varieties and storage temperature for progeny emergence, percentage of insect damaged kernels, grain weight loss, and amount of powder produced per gram of wheat. Kernels of Danda'a, infested with S. oryzae at 30 °C exhibited significantly lower mean progeny counts (13.3 live insects), lower percentage of grain weight loss (4.2%) and insect-damaged kernels (6.4%), and powder production (1.5 mg/g). Kernel weight and hardness index were negatively associated with percentage of insect damaged kernels and grain weight loss. Kernel diameter was positively associated with both of percentage of insect damaged kernels and grain weight loss. Wheat varieties with high Zeleny sedimentation values had lower percentage of insect-damaged kernels and grain weight loss. These results indicated that kernel weight, hardness index, and protein content are predominant factors contributing to wheat resistance against S. granarius and S. oryzae. The varieties Millennium and Danda'a can be considered with other integrated pest management approaches to reduce stored grain losses of wheat in Ethiopia. 相似文献
12.
13.
王永 《中国印刷与包装研究》2014,(2):14-19
为了降低图像分类算法的计算复杂度,提高图像分类的准确性,本研究提出一种基于稀疏非负张量分解的图像分类算法,首先提取图像本身的结构特征信息得到图像特征数据,再把子空间数据稀疏性作为约束项,添加到非负张量分解目标函数中,再利用稀疏约束的非负张量分解算法对图像数据集进行降维处理,最后使用支持向量机方法对图像数据库进行分类。实验结果表明,本研究提出的算法能有效提高图像分类的准确性并降低计算复杂度。 相似文献
14.
The NIS‐Elements image analysis system, together with microscopic observation, was used to measure the particle size of acetylated distarch adipate (ADA) in three suspensions; a commercial mixture containing other additives and two pure commercial ADA starches. From these data, the particle size distribution was calculated. The starch in the mixture was identified as wheat ADA, and the normality of the size distribution for volume fraction tested. Image analysis can be recommended as a simple and rapid method for identifying the origin of modified starches in commercial powder mixtures. 相似文献
15.
为了沁州黄小米的产地溯源和品牌保护,本研究采集沁州黄小米地理标志保护区及域外小米样品,利用计算机图像处理技术对来源于沁州黄小米保护区(核心区和非核心区)和域外区3个产区的小米有效颜色特征进行提取。分别建立Fisher判别、K最近邻判别和BP人工神经网络3种判别模型并进行比较。采用紫外可见光谱法和荧光定量PCR法对不同产区小米的总类胡萝卜素含量及其降解途径中的关键基因进行差异分析。结果表明:不同产区的沁州黄小米在RGB色彩模式中B值存在显著差异(P<0.05);3种判别模型中,BP人工神经网络判别模型正确率最高为96.7%,具有可行性。进一步分析表明,小米颜色参数变化与总类胡萝卜素含量有显著相关性,并且不同产区之间总类胡萝卜素含量存在显著差异(P<0.05),核心产区小米的总类胡萝卜素含量最高,达到15.89 mg/kg;此外,沁州黄核心区小米中的类胡萝卜素裂解双加氧酶的表达量显著低于非核心区和域外区(P<0.05)。本研究运用计算机图像处理技术和人工神经网络相结合的判别方法对沁州黄小米的不同产地进行了正确区分,具有一定的应用价值,可为沁州黄小米的产地溯源和品牌保护提供技术支持,同时对类胡萝卜素含量及降解途径中关键基因进行比较分析,从生化和分子水平揭示了引起不同产区沁州黄小米图像和品质差异的本质原因。 相似文献
16.
玉米品种图像识别中的影响因素研究 总被引:1,自引:1,他引:1
为了研究玉米品种图像识别中的关键影响因素,搭建了一套基于PCA和ICA特征提取和支持向量机(SVM)分类算法的玉米品种识别系统,采用扫描仪获得了11个品种每个品种50粒图像,基于图像的像素特征和统计特征,分别研究了主分量分析(PCA)和独立分量分析(ICA)的特征提取和特征优化方法,并进一步考察了支持向量机(SVM)模式分类过程中的关键参数优化问题.试验结果表明,对11个品种550个籽粒的品种最高检出率为97.17%,在同样的情况下ICA优化的特征较PCA优化的特征识别率能提高3%左右,适当选择统计特征比使用像素特征识别率提高约10%,另外SVM参数影响到识别效果,但整体影响不大.本方法与结论对玉米种子纯度和品种真实性检验具有积极意义. 相似文献
17.
Martyn W Griffiths Andrew D E Martin Trevor J Hocking Steve B Reynolds 《Journal of the science of food and agriculture》1990,51(1):27-33
The accurate analysis of sulphur and sulphur compounds in plant material has become a critical issue as a consequence of reduced availability of the element from the environment. A new application of a reproducible technique has been developed for the determination of sulphur in wheat grain using X-ray fluorescence spectrometry. This new application compares favourably with traditional wet chemical methods especially as it is non-destructive. The sulphur content in wheut yruin measured by X-ray ffuorescence spectrometry was found to have a lower variability and the speed of analysis was more rapid in comparison with the chemical method, enabling a faster throughput of samples. Because sample preparation is very important for accurate analysis, the problems associated with oven drying, milling and of pressing during pellet formation are discussed. 相似文献
18.
Jensen TH Böttiger A Bech M Zanette I Weitkamp T Rutishauser S David C Reznikova E Mohr J Christensen LB Olsen EV Feidenhans'l R Pfeiffer F 《Meat science》2011,88(3):379-383
X-ray computed tomography (CT) has recently received increased attention in the food science community. The aim of this paper is to demonstrate how grating based phase-contrast CT can provide contrast superior to standard absorption based CT. The method of phase-contrast CT is applied to two samples of porcine subcutaneous fat and rind. The additional contrast obtained may be used for quality testing, to investigate variations in fatty acid composition of the fat-fraction, and density variations in the meat-fraction. The possibility of integrating the method into an abattoir environment is discussed. 相似文献
19.
种子是农业生产最基本,最主要的生产资料。为实现玉米种子的快速鉴定识别与保护,本文提出基于卷积神经网络(Convolution Neural Network, CNN)与迁移学习相结合的玉米种子籽粒图像分类识别方法,可将预训练的CNN模型参数迁移到玉米籽粒图像分类识别任务中。实验采集了6种玉米种子籽粒图像双面图像共1 976张,包括16DX531,京粘1号,科诺58,铁研,小金黄,郑单958,建立胚面,胚乳面和双面混合的3种数据集。按照7∶2∶1的比例随机划分训练集,验证集和测试集,并对训练集图像作数据增强处理。基于4种CNN模型Xception, ResNet50V2,MobileNetV2,DenseNet121进行参数迁移学习。实验结果表明Xception与胚乳数据集建模方法优于其他方法。Xception--胚乳模型训练集与验证集平均识别准确率分别为95.55%和95.97%,测试集准确率为92.78%。基于卷积神经网络与迁移学习相结合的玉米籽粒图像识别方法切实可行。 相似文献