基于MATLAB图像处理的大豆颗粒检测方法研究

为解决大豆单株考种过程中人力计数准确率低和数粒仪等光电方法耗时长等问题,采用计算机视觉系统及MATLAB软件开发平台代替人工大豆单株考种进行自动检测的方法。该算法采用对大豆颗粒图像进行空间滤波去除噪声及“Otsu”方法对图像进行最佳全局阈值分割处理,在对图像处理的基础上完成对大豆颗粒个数、颗粒大小两项指标的测定。以六个大豆品种周黑豆、周青豆、周豆11、周豆18、周豆22和周豆23的籽粒为试验对象,探究大豆颗粒计数、大豆颗粒大小分级两项指标的确定。实验结果表明：该算法及程序准确有效,能准确计算大豆单株颗粒个数;该算法及程序判断出的各品种大豆颗粒平均面积大小与百粒重成正相关,且决定系数为0.9141,即可得该算法及程序可准确有效的判断大豆颗粒大小。总之,基于MTALAB图像处理大豆单株颗粒检测方法的研究可相对减轻人力劳动强度及人类视觉的不足,在提高工作效率和准确度等方面有重要意义。     

基于GWO优化SVM的小麦籽粒优劣分级研究

小麦籽粒优劣不仅是产量及品质的重要决定因素,也是育种适应性的综合指标。为了提高小麦籽粒优劣分级的准确率,同时克服神经网络中存在的收敛速度慢、容易陷入局部极值等缺陷,提出一种灰狼算法（GWO）优化支持向量机（SVM）的小麦籽粒优劣分级方法,以航麦8805为研究对象,利用图像处理技术对小麦籽粒图像进行预处理并提取小麦籽粒的形态、颜色和纹理等21个特征。然后采用灰狼算法对支持向量机的两个参数（c、σ）进行优化,建立GWO-SVM模型,从而对小麦籽粒进行优劣分级。与其他算法相比,GWO优化SVM的算法对小麦籽粒的分级准确率有明显的提高,对小麦籽粒优劣分级的准确率可达到95.08%。     

基于近红外光谱的小麦品质分类研究

为了快速、简便、准确地鉴别小麦品质的类别,本研究提出了应用近红外光谱分析技术结合BP神经网络的鉴别方法对小麦进行品质分类。研究过程中对小麦样品的光谱数据进行了详细分析,采用马氏距离剔除了光谱数据中异常数据,并通过主成分分析说明利用近红外光谱鉴别小麦品质分类的可行性。为了提高所建模型的性能,采用SPXY算法对小麦样品进行合理的划分。并选取了一阶微分加归一化的预处理方法来处理光谱数据,消除无关信息和噪声对小麦光谱数据的影响。运用偏最小二乘法压缩光谱数据,减少了数据量,节省建模时间。最后采用BP神经网络方法建立了小麦品质分类模型。实验结果显示：模型的鉴别效果较好,对强筋样品识别的准确率高达94.4%,弱筋样品识别的准确率高达100%。实现了快速、准确地对小麦品质强筋和弱筋两类的鉴别,对小麦生产、市场交易及食品加工有着非常重要的意义。     

Research and implementation on handwritten numeralrecognition technique based on fuzzy mode

Based on simulation and numeral recognition of the 24-bit bmp images of postal codes obtained by CCD transducer, a binarization method based on coloring matter is proposed, which is employed to extract the binary image of the frame line of postal code and the binary image of the mail character separately. The procedure of image processing is analyzed, which includes detecting and separating borderlines of postal code frames, image binarization, smoothing noise-removal, slant rectification, extraction of postal code numerals, position normalization processing, subdivision of digital image, feature extraction of character recognition, and so on. The pretreatment process of image recognition is described. Then, character feature extraction is conducted for pretreated images, while the angle correctness for postal code frames and characters is performed. Finally, the recognition result is determined. In this method, the difficulty of separating frame line from postal code character line is eliminated. In order to improve recognition rate, the combination of many methods of handwritten numeral is used, resulting in improved recognition rate of postal codes. The experiment shows that this method can obtain higher recognition rate, providing reference for the separation technique in postal system.     

基于多光谱成像技术的小麦品种快速无损鉴定

为了研究多光谱成像技术对小麦品种快速无损鉴定的可行性,采用VideometerLab 多光谱图像采集设备对5 个小麦品种共500 个样品在405~970 nm波段内的进行多光谱图像信息进行采集,获取其光谱、颜色和形态特征。利用主成分分析对5 个小麦品种进行定性鉴别,同时,基于光谱特征和光谱图像特征分别比较了神经网络、支持向量机和随机森林3 种模型的鉴定效果。结果显示：利用19 个光谱特征值建立的模型中,BPNN识别模型效果最佳,其建模集和预测集的识别率分别为100%和91.25%。融合19 个光谱特征和6 个图像特征所建立的模型中,BPNN识别模型效果最佳,其建模集和预测集的识别率分别达到了100%和98.4%。结果表明,基于BPNN的多光谱特征融合能够有效的提高小麦品种鉴定效率,为小麦品种的快速无损检测提供了一个新途径。     

基于粒子群算法和支持向量机的黄花菜叶部病害识别

使用数字图像处理技术,以黄花菜叶部病害图像为识别对象,基于Lab空间和K-means聚类算法分割病害区域,提取目标区域的颜色特征、方向梯度直方图(histogram of oriented gradient,HOG)特征和形状特征,分别建立单一特征模型和特征融合模型,采用粒子群(particle swarm optimization,PSO)算法通过交叉验证优化支持向量机(support vector machine,SVM)模型的惩罚因子和核参数,建立基于PSO-SVM的多特征融合分类模型识别黄花菜病害。基于SVM的多特征融合分类模型识别率高于单一特征分类模型,识别率可达为81.67%;基于PSO-SVM多特征融合分类模型识别率高达92.39%。基于PSO-SVM的多特征分类模型识别率高,可以及时、便捷、高效地识别黄花菜病害。     

Design of Vision System Based on Service Robot

A vision system based on the service robot is involved. In the system, picture signal is acquired by the picture sensor OV7635. Frame memory AL422B is used as data buffer memory, while CPLD controlled the time order DSP performed. In the software system of image processing, to accomplish color image segmentation and recognition, the threshold vector judgment and improved seed-fill algorithm is introduced, and the image geometric moment is calculated during the segmenting. In order to achieve a vision servo system which composed of image-based feedback and adaptive compensation, the deduced matrix-based Jacobian from the image moment is taken as image feature. Adopted the TFT LCD is adopted to straightly disolav the result of vision recognition and vision tracing.     

Technique of Surface Reconstruction for Three Dimensional Structure of Chinese Medicihnal Materials

To study the technique of surface reconstruction for Chinese medicinal materials from its cross-sectional images and to display its three dimensional structure.The pattern recognition technique is used to complete the classification and matching of various structures, then the triangle surface approximation is used to produce the 3D surface of the structure.The 3D structure is reconstructed and can be viewed by flexible way based on SGI workstation. The technique and method provided by this paper is contributive to the study and view of various structure of Chinese medicinal materials, and is available to biomedical image processing field.     

Design of nondestructive test system for plasma clots

One of the most important measures that are used to guarantee blood transfusion safety is to detect clots in the plasma before transfusion. To overcome the disadvantages of manual detection method, this research designs a nondestructive testing (NDT) system for plasma clots inspection based on machine vision technique and artificial neural networks. The key technology for system design are studied and presented. Image acquisition is performed by custom-designed software based on MATLAB platform, and the methods of image cut, reverse color, median filter as well as gray cutting are adopted to preprocess image. The use of fisher discrimination method, combined with iterative threshold segmentation method and the selection of connected domain, can successfully eliminate the interference of air bubble and correctly extract the image of plasma clots. Plasma clots are discriminated by a recognition model based on artificial neural network BP algorithms. The results of clinical contrast experiment shows that the system can effectively detect whether plasma contains plasma clots and the new system shows a much higher degree of repeatability and stability. From the image acquisition and processing to the recognition of plasma clots, the detecting time of a sample is no more than 1 min.     

基于数字图像的玉米叶面积测量方法研究

【研究目的】以数字图像处理技术为基础,采用数码相机获取数字图像,建立使用Visual C++6.0开发的图像处理软件进行玉米叶面积测量的方法。【方法】首先,用数码相机拍摄挂在墙上粘贴着玉米叶片的参照白板来获得数字图像。然后提取图像的蓝色分量,利用中值滤波、膨胀、腐蚀等方法去除图像噪声,最大类间方差法自动阈值分割目标。接着进行区域标记与区域像素数统计,区域像素数结合图像分辨率就可以获得玉米叶片的实际面积。【结果】与直尺法所测结果具有很好的相关性,且比直尺法更精确,【结论】此种拍摄方法避免了几何畸变的产生。蓝色分量图像中目标和背景的灰度差异最大,便于分割。     

不同用途优质小麦籽粒品质、产量与环境效应基因型差异研究

不同优质小麦因其具体用途不同,对蛋白质含量、面筋含量、沉降值等指标要求不同。通过3种不同用途的优质小麦品种在9种不同水肥处理条件下的产量和籽粒品质的研究,结果表明：不同的基因型品种其籽粒品质和产量指标存在极显著差异;小麦产量及籽粒品质主要决定基因遗传效应,同时水肥效应和品种水肥互作效应对籽粒品质和产量有极显著影响。不同基因型优质小麦籽粒品质和产量指标在水肥处理间变异程度不同。     

机器视觉检测鲜烟叶的分级装置设计

针对农民在烟叶送烤前对烟叶分级的非重视度和非客观性等问题,本文提出基于机器视觉技术的烟叶图像检测分类方法对烟叶编烟送烤前进行成熟度划分,设计了全自动化的鲜烟叶检测分级装置。该系统的机械结构由自动上样抓取烟叶机构、烟叶输送台、检测机构,分拣机构等四个部分组成。自动上样的烟叶在传送带上被CCD检测并进行图像处理。首先对图像的噪点采用邻域平均和中值滤波组合的方法进行区域去噪处理;使用最小误差阈值分割方法分离背景和烟叶,然后增强图像信号,提取感性区域的颜色信息。采用烟叶的4个特征信息(R,G,B颜色值和色调H值)来表征烟叶的级别特性。通过装置自主学习建立样本库,然后参考学习的样本库对未知样品进行检测分级。实验结果表明根据烟叶色泽的不同,相邻类型之间的色泽差异越大,分类准确度越高。检测分类的平均速度在2-3秒/片,满足现场即时检测要求。     

Development of Data Acquisition and Processing System Based on DSP and USB 2.0

The data acquisition and processing system are proceeding towards faster, more powerful, easier to use and more cost effectiveness. This paper discuss the design of the high speed and convenient data acquisition and processing system based on DSP and USB 2.0, and expatiates the principle and method of the virtual instruments system in detail. The system employs an ADS8346 to sample the six channel signals simultaneously with a resolution of 16-bit.It utilizes EZ-USB FX2 as the USB2.0 interface, which can achieve the high speed communication between the host and the system. It uses TMS320VC33 (dsp) as the core chip, which can accomplish the function of the main control and real-time digital signal processing. The system enhances the virtual instruments' advantage more by the combination between software and hardware.     

应用15N研究小麦各部位氮素分配利用及施肥效应

应用同位素示踪技术通过对盆栽小麦的不同施肥处理,研究小麦不同部位对氮素肥料的分配利用规律,结果表明,小麦主茎穗籽粒的ＮＤＦＦ为２８．５９％,分穗籽粒的ＮＤＦＦ为２９．３０％,茎秆的ＮＤＦＦ为２６．０２％,根系的ＮＤＦＦ为３２．１６％,而土壤的ＮＤＦＦ仅为１．８８％。从氮肥的利用率分析,籽粒的氮肥利用率为１９．５５２％。茎秆的利用率为３．４９０％,根系的利用率为０．６７４％,氮肥的回收率平均为４１．     

应用数字图像技术估测冬小麦冠层生物量垂直分布特征的研究

用数字图像技术研究了冬小麦冠层生物量的垂直分布。表明用一行小麦图像比多行小麦图像估测小麦生物量能更好地满足线性回归关系,估测效果更佳,以此为基础进一步研究了分层像素数估测小麦冠层分层现存生物量和有效生物量的方法。利用分层绿色像素数(LGPN)指标定性分析了不同栽培模式下冬小麦群体有效生物量的垂直分布和动态变化,并确定了基于图像特征的可用于定量分析的小麦群体垂直分布指数（I）。     

麦麸保鲜面的工艺与保鲜技术研究

研究了麦麸保鲜面的配方及其保鲜工艺技术。通过对比麦麸保鲜面水分含量、环境温湿度、原料粗细度、微生物控制等与保质期、产品品质的关系,确定了麦麸保鲜面的最佳配方和工艺参数。     

遥感影像分类结果的不确定性研究

主要从遥感数据质量的不确定性、分类算法的不确定性、分类评价方法的不确定性及分类后处理的不确定性四个方面讨论了遥感影像分类结果的不确定性。     

挤压膨化处理麸皮对全麦粉品质的影响

将红麦、白麦、加麦的小麦麸皮经挤压膨化处理、粉碎后,按原配比回添到面粉中,制得全麦粉。通过与未处理小麦麸皮回添的全麦粉在理化指标和流变学特性对比来评判2种工艺制得的全麦粉的性质,从色度、面筋、蛋白、沉降值、粉质、拉伸等方面进行比较,挤压膨化处理麸皮回添工艺对全麦粉的品质有明显的改变。     

Study on ANN Based Character Recognition

The theory of Artificial Neural Networks and image processing are analyzed. ANN theory is also investigated . A Back Propagation network is used for classification. Based on the ANN and feature extraction, a character recognition and entry system is developed. Its drawbacks are discussed. Some solutions are presented. An odd feature database is constructed to promote the performance for learning. A 3 layer feedforward network has been proved that it has the ability to identify hand written characters.     

施氮时期对小麦籽粒HMW-GS积累及GMP粒度分布的影响

在225 kg hm^-2施氮水平下, 设置起身肥(SE, GS 30), 拔节肥(JT, GS 32)和孕穗肥(BT, GS 41) 3个追施氮肥处理(底追比1:1), 研究了追肥时期对强筋小麦济南17和弱筋小麦鲁麦21籽粒HMW-GS积累和GMP粒度分布的影响。结果表明, 两品种籽粒HMW-GS于花后14 d均已形成, 济南17籽粒HMW-GS和GMP含量均高于鲁麦21, 说明强筋小麦具有较强的谷蛋白积累能力。济南17成熟期籽粒HMW-GS和GMP含量以SE处理最高, 施氮时期后移其含量呈下降趋势。JT处理显著提高鲁麦21灌浆中后期HMW-GS的积累速度, 延长HMW-GS的快速积累期。济南17 SE处理和鲁麦21 JT处理的籽粒x型亚基(1、4或5、7)在灌浆中后期的积累速率显著提高。追肥时期对y型亚基的积累速率无显著影响。追氮时期后移均提高两品种籽粒GMP小颗粒的(粒径<12 μm)体积和表面积百分比, 降低大颗粒(粒径>100 μm)体积和表面积百分比。济南17粒径>12 μm的GMP颗粒数目百分比因追氮时期后移而增加, 鲁麦21粒径>12 μm的GMP颗粒数目百分比则降低。含4+12亚基的强筋小麦济南17比含5+10亚基的弱筋小麦鲁麦21偏向于更高的大颗粒体积比例, 说明亚基间的聚合和GMP颗粒的分布不仅与亚基类型有关, 而且与单位面粉中亚基的含量密切相关。