大田害虫实时检测装置及其图像预处理技术的研究
|
摘要
我国目前正在农业领域大力推广IPM(农业有害生物的综合治理)工程。IPM的主要目标是一方面利用各种防治方法尽可能地减少有害生物造成的损失；另一方面尽可能地减小投入环境的化学农药，最终应能保障乃至增强农业的可持续发展。由此可见，农业病虫害检测预报工作是IPM工程的重要组成部分，它也是各级植保部门的主要工作。
目前普遍采用的黑光灯诱集害虫、人工识别计数的测报方法，存在识别准确性差、效率低等严重缺陷，极大地降低了测报的准确度和时效性，不利于农田害虫的防治工作及时、有效、有针对性地展开，因此本文提出了基于图像工程和计算机视觉的图像识别技术应用于农田害虫的自动检测预报。
本文设计制作了诱捕装置，诱集农田害虫，使用摄像头摄取害虫图像，并在此基础上进行了有针对性的图像预处理技术的研究；通过对所获取目标图像形态学特征的提取和选择，提出了利用害虫形态学特征进行种类识别的观点，将压缩后特征输入神经网络分类器进行识别，给出检测结果。验证了获取目标图像的装置和应用图像处理的方法的可行性和有效性。
本文针对农业大田害虫这一研究对象，进行了以下工作：
1．诱集和传输装置设计。采用植保站常用的20W、3600A黑光灯诱集害虫，设计装置使害虫落入流动的水中，借此调整害虫的姿态和保证传输过程的均匀稳定。
2．均匀光照系统的设计。借鉴无影灯原理，采用对称光源和漫反射材料制作均匀光照室保证获取图像的光照质量。
3．图像采集接口的设计。充分利用图像卡提供的良好性能，基于OK—C30S卡的API进行二次开发，利用其回调功能实现采样和处理的并行完成。
4．对图像进行预处理。在对图像进行灰度化、二值化的基础上，利用各种优化的处理算法实现图像增强处理，消除噪声，分割出目标。
5．特征选择和提取。这是对害虫图像进行识别的关键环节，分别提取了面积、周长、复杂度、七个不变矩等十四个特征，形成原始特征空间；利用特征压缩也即特征选择和提取的方法对原始特征空间压缩进行压缩。
6．农田害虫的识别与分类。
与研制的硬件装置相配合，用Visual C++ 6.0语言和Matlab图像处理工具箱实
郑州大学硕士学位论文
现了上述各环节的功能。另外还在VisualC++平台上调用SQL Server实现了用户
管理、数据管理、数据传送模块的开发，从而构建了一个较为完整的图像识别检
测系统。
由于本人的水平有限和一些客观条件的限制，论文试验在软硬件设计上还存在
很多有待改进的地方。不过本人相信图像自动识别系统具有广阔的发展前景。
Nowadays, China is pushing IPM greatly in agriculture. One aim of project is to decrease the amount of lost resulted from pest by using various ways. Another is to reduce the use of pesticide in environment as possible in order to ensure and even strengthen the sustainable development of agriculture. Therefore, insect pest detection and prediction of agriculture are important part of IPM project. It is the groundwork of pest control services at all levels too.
At the present time the prediction method of attracting pest with black light and recognizing and counting by man is adopted generally. There are some serious shortages such as bad recognition accuracy and low efficiency. It reduces seriously accuracy and timeliness of prediction and is disadvantage in guiding insect disease prevention. Therefore, this paper researched image recognition technology based on image processing. This technology will automatically detect and predict the amount and sort of various insect pests.
We designed and made the attraction device to attract agricultural pests, obtained agricultural pests' images with the color camera and processed images . On the basis of these, we emphasized on extracting effective features based on the theory of Mathematics morphology. Then we selected features, inputted into neural network classifier, recognition, presented detection results.
For this research object, pest of the field in agriculture, the paper carry on the following work:
1. To design the attraction and transit device. We attracted insect pests with black light which was used by pest control services, adjusting insect pests' posture by running water to ensure a steady transit.
2. To offer equality and invariable lamp. Using the theory of shadowless lamp, we made even illumination room guarantee illumination quality to obtain the image through adopting symmetrical light source and diffuse reflection material.
3. To set image collection mode by programming based on the OK-C30S image card API. It realized parallel running of the image recognition system.
4. The processing of pests' image. In this tech-develop stage, We fulfilled images enhancement The grain-pests image is smoothed by the gray morphology, enhanced by the adaptive method. Eventually, we segmented the object images form their
background.
5. Feature extracting and selecting. It is the crucial part of the system. What features we selected are as follow: area, perimeter, complexity, equivalent circle radius, and so on. Then, we made use of K-L translation based on the global in-class discrete matrix and the distance separable rule to realize the features' compressing.
6. Image recognition and classifier design.
We implemented mentioned above functions with Visual C++6.0 language, developed software package, associated with designed hardware system.
Due to my poor tech and other confining of some objective condition, there are still a lot of shortcomings in design of hardware and software. However, I firmly believe that a desirable future in this field lies open to us.
目前普遍采用的黑光灯诱集害虫、人工识别计数的测报方法，存在识别准确性差、效率低等严重缺陷，极大地降低了测报的准确度和时效性，不利于农田害虫的防治工作及时、有效、有针对性地展开，因此本文提出了基于图像工程和计算机视觉的图像识别技术应用于农田害虫的自动检测预报。
本文设计制作了诱捕装置，诱集农田害虫，使用摄像头摄取害虫图像，并在此基础上进行了有针对性的图像预处理技术的研究；通过对所获取目标图像形态学特征的提取和选择，提出了利用害虫形态学特征进行种类识别的观点，将压缩后特征输入神经网络分类器进行识别，给出检测结果。验证了获取目标图像的装置和应用图像处理的方法的可行性和有效性。
本文针对农业大田害虫这一研究对象，进行了以下工作：
1．诱集和传输装置设计。采用植保站常用的20W、3600A黑光灯诱集害虫，设计装置使害虫落入流动的水中，借此调整害虫的姿态和保证传输过程的均匀稳定。
2．均匀光照系统的设计。借鉴无影灯原理，采用对称光源和漫反射材料制作均匀光照室保证获取图像的光照质量。
3．图像采集接口的设计。充分利用图像卡提供的良好性能，基于OK—C30S卡的API进行二次开发，利用其回调功能实现采样和处理的并行完成。
4．对图像进行预处理。在对图像进行灰度化、二值化的基础上，利用各种优化的处理算法实现图像增强处理，消除噪声，分割出目标。
5．特征选择和提取。这是对害虫图像进行识别的关键环节，分别提取了面积、周长、复杂度、七个不变矩等十四个特征，形成原始特征空间；利用特征压缩也即特征选择和提取的方法对原始特征空间压缩进行压缩。
6．农田害虫的识别与分类。
与研制的硬件装置相配合，用Visual C++ 6.0语言和Matlab图像处理工具箱实
郑州大学硕士学位论文
现了上述各环节的功能。另外还在VisualC++平台上调用SQL Server实现了用户
管理、数据管理、数据传送模块的开发，从而构建了一个较为完整的图像识别检
测系统。
由于本人的水平有限和一些客观条件的限制，论文试验在软硬件设计上还存在
很多有待改进的地方。不过本人相信图像自动识别系统具有广阔的发展前景。
Nowadays, China is pushing IPM greatly in agriculture. One aim of project is to decrease the amount of lost resulted from pest by using various ways. Another is to reduce the use of pesticide in environment as possible in order to ensure and even strengthen the sustainable development of agriculture. Therefore, insect pest detection and prediction of agriculture are important part of IPM project. It is the groundwork of pest control services at all levels too.
At the present time the prediction method of attracting pest with black light and recognizing and counting by man is adopted generally. There are some serious shortages such as bad recognition accuracy and low efficiency. It reduces seriously accuracy and timeliness of prediction and is disadvantage in guiding insect disease prevention. Therefore, this paper researched image recognition technology based on image processing. This technology will automatically detect and predict the amount and sort of various insect pests.
We designed and made the attraction device to attract agricultural pests, obtained agricultural pests' images with the color camera and processed images . On the basis of these, we emphasized on extracting effective features based on the theory of Mathematics morphology. Then we selected features, inputted into neural network classifier, recognition, presented detection results.
For this research object, pest of the field in agriculture, the paper carry on the following work:
1. To design the attraction and transit device. We attracted insect pests with black light which was used by pest control services, adjusting insect pests' posture by running water to ensure a steady transit.
2. To offer equality and invariable lamp. Using the theory of shadowless lamp, we made even illumination room guarantee illumination quality to obtain the image through adopting symmetrical light source and diffuse reflection material.
3. To set image collection mode by programming based on the OK-C30S image card API. It realized parallel running of the image recognition system.
4. The processing of pests' image. In this tech-develop stage, We fulfilled images enhancement The grain-pests image is smoothed by the gray morphology, enhanced by the adaptive method. Eventually, we segmented the object images form their
background.
5. Feature extracting and selecting. It is the crucial part of the system. What features we selected are as follow: area, perimeter, complexity, equivalent circle radius, and so on. Then, we made use of K-L translation based on the global in-class discrete matrix and the distance separable rule to realize the features' compressing.
6. Image recognition and classifier design.
We implemented mentioned above functions with Visual C++6.0 language, developed software package, associated with designed hardware system.
Due to my poor tech and other confining of some objective condition, there are still a lot of shortcomings in design of hardware and software. However, I firmly believe that a desirable future in this field lies open to us.
引文
[1]沈佐锐．现代信息技术的发展对持续植保的影响．中国植物保护研究进展．中国科学技术出版社1996：73-77
[2]张家兴．可持续农业中21世纪初的IPM．植物保护．1998,24(6)：38～40
[3]Pursley W.Insect electrocutor light traps.AIB Tech.Bull.1987,9:1～5
[4]Huang J X.Synthesis of the stereoisomers of the sex pheromones of the southern corn rootworm.Chemical Research in Chinese.2001, 17(1): 51～55
[5]陈庭华．斜纹夜蛾大发生原因和性诱剂的诱蛾效果．长江蔬菜．1997(7)：22～24
[6]许建．害虫成虫防治技术及应用展望．新疆农业大学学报．2000，23(S1)：32～34
[7]Robin M R.Sticky trap for monitoring leafin iners Liriomyza sativae and Liriomyza trifolii (Diptera; Agromyzidae) and their associated bymenopterous parasites in watermelon.J.of Econ.Entomol.1987,80(6): 1345～1347
[8]何忠．蝼蛄的声学通讯．生物学通报．1995,30(12)：6～7
[9]Vick, K.W..Sound detection of stored-product insects that feed inside kernels of grain.Econ.Entomol.1988,81:1489～1493
[10]Menne D.Short-time spectral analysis of the ultrasound echoes of flying insects.Localization and Orientation in Biology and Engineering,Springer-Verlag.1984:225～230
[11]沈佐锐，于新文．昆虫数学形态学研究及其应用展望．昆虫学报．1998,41(S1)：140～148
[12]于新文，沈佐锐．昆虫数字图像的分割技术研究．农业工程学报．2001，17(3)：137～141
[13]王秀丽译．昆虫学家发明害虫监听系统．世界农业．2001，11(总271)：59～63
[14]Jordan, Jay B,.Vision guided insect handling system, Modeling and Simulation.Proceedings of the Annual Pittsburgh Conference.1990, 21(5): 1995～2000
[15]Subramanyam, Bh..Alternatives to pesticides in store-prouction IPM.Kluwer Academic Publisher.2000:58～65
[16]赵志模．嗜卷书虱的试验生态研究．昆虫学报．1999,42(3)：277～282
[17]张红涛．储粮害虫图像识别中的特征抽取研究[硕士学位论文]．郑州大学．2002．
[18]张法全．基于机器视觉和小波分析的农田害虫图像识别系统[硕士学位论文]．郑州大学．2003．
[19]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究．Ⅰ．在目级阶元上的应用研究昆虫学报2003 vol46(1)45-50．
[20]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究．Ⅱ．在总科阶元上的应用研究昆虫学报2003 vol46(2)201-208．
[21]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究Ⅲ．在科阶元上的应用研究 昆虫学报2003 vol46(3)339—344．
[22]于思勤，孙元峰．河南农业昆虫志．北京：中国农业科技出版社．1993
[23]朱志刚，林学訚，石定机等．数字图像处理．北京：电子工业出版社．2002
[24]章毓晋．图像工程(上册)——图像处理和分析．北京：清华大学出版社．2000
[25]章毓晋．图像工程(下册)——图像理解与计算机视觉．北京：清华大学出版社．2000
[26]田村秀行[日]．计算机图像处理技术．北京：北京师范大学出版社，1986
[27]何斌数字图像处理北京：人民邮电出版社2003
[28]崔屹．数字图像处理技术与应用．北京：电子工业出版社．1997：187～200
[29]李刚，范瑞霞一种改进的图像中值滤波算法北京理工大学学报2002，(6)376～378
[30]雷晓峰，王耀南，邓曙光．在VC++中利用多线程实现图像采集与通信微型机与应用2002，2：6～8
[31]杨威，张田文，师海峰．一种用于二值图像分割的快速聚类算法计算机研究与发展 1998，35(8)：719～723
[32]高雁飞，范九伦．一种新的灰度图像阈值分割方法西安邮电学院学报．2002，7(1)：1～4
[33]石俊生，杨卫平，白凤翔，张韬，唐伟．彩色图像分割与颜色特征选择，云南师范大学学报1999,19(6)：53～56
[34]刘志敏，杨杰．基于数学形态学的图像形态滤波红外与激光工程1999，28(4)：10～15
[35]Running, S.W., Loveland, T.R., Pierce, L.L., et al.A Remote Sensing Based Vegetation Classification Logic for Global LandCover Analysis.Remote Sensing of Environment, 1995, 51: 39-48.
[36]M.Egmont-Petersen,D.de Ridder, H.Handels Image processing with neural networks-a review.Pattern Recognition 35 (2002) 2279-2301
[37]Mario K?ppena, Javier Ruiz-del-Solar.Steady-state image processing.Applied Soft Computing 1(2001) 53～62
[38]Robert Sablatnig, Martin Kampel.Model-Based Registration of Front- and Backviews of Ro- tationally Symmetric Objects.Computer Vision and Image Understanding 87 (2002), 90-103
[39]Qing Wang, Zheru Chi, Rongchun Zhao.Image Thresholding by Maximizing the Index of Non- fuzziness of the 2-D Grayscale Histogram Computer Vision and Image Understanding 85(2002) 100-116
[40]贾振堂，李生平，贺贵明等一种基于运动边缘检测的视频对象分割新算法计算机研究与发展2003(5)：684-689
[41]吕同富，刘宝军，毕秀芝图像边缘提取的简单方法及应用计算机仿真2003 20(4)：99-101
[42]张爱华，余胜生，周敬利．一种基于边缘检测的局部阈值分割算法小型微型计算机系统 2003 24(4)：661-663
[43]Changick Kim, JenqNeng Hwang.A fast and robust moving object segment in video sequences IEEE Computer Society,1999 131～134
[44]Kevin Bowyer, Christine Kranenburg, Sean Dougherty Edge Detector Evaluation Using Empirical ROC Curves Computer Vision and Image Understanding 2001 (84), 77-103
[45]谈正，刘辉，周军．旋转及伸缩不变纹理分割算法 西安交通大学学报。1997，31(1)：32～37
[46]Yamashita, Y., 0gawa, H.Relative Karhunen-Loeve transform.IEEE Tans.on Signal Processing, 1996, 44 (2): 371-378.
[47]Hua, Y.Liu, W.Generalized Karhunen-Loeve transform.IEEE Signal Processing Letters, 1998, 5(6): 141-142.
[48]江铭虎，季文铎，林碧琴等．一种特征压缩及分类神经网络的研究．电路于系统学报，1997，2(3)：18-23．
[49]Kung, S.Y.Adaptive principal component extraction(APEX) and application.IEEE Trans.Signal Processing, 1994, 42 (5): 1202-1217.
[50]Defries, R.S., Townshend, J.R.G.NDVI-derived L and Cover Classification data.Global Scale.International Journal of Re-mote Sensing, 1994, 15 (17): 3567-3586.
[51]邱道尹，张成花，张红涛等，神经网络在储粮害虫识别中的应用，农业工程学报
[52]唐向宏，贺振华，杨绍国．利用小波变换进行地震数据压缩．成都理工学院学报，1999，26(2)：183-186．
[53]张永胜，郁可．基于小波神经网络的光谱数据压缩与分类研究．计算机研究与发展，1999，36(8)：973-977．
[54]向阳，龚新高．推广模拟退火方法及其应用．物理学进展，2000，20(3)：319-334．
[55]Sidelecki, W., Slansky, J.A note on genetic algorithms for large-scale feature selection.1989,10 (5): 335-347.
[56]Roberto, B.Using mutual information for selection features in supervised neural net learning.IEEE Trans.on NN, 1994, 5 (4): 537-550.
[57]边肇祺，张学工．模式识别(第二版)．北京：清华大学出版社，2002．
[58]金连文，徐秉铮．基于多级神经网络结构的手写体汉字识别．通信学报，1997，18(5)：21-27．
[59]张学工．关于统计学习理论与支持向量机．自动化学报，2000，26(1)：32-42．
[60]Widrow B, et al.Neural networks application in industry, business and science Communication of the ACM.1994,37:93～105.
[61]高大启．有教师地线性基本函数前向三层神经网络结构研究．计算机学报．1998，21(1)80～86．
[62]李平，刘炯．前馈神经网络的结构与范化，长春光学机械学院学报．1997，20(2)：63～67
[63]阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社．2000：45～78
[64]黄德双．神经网络模式识别系统理论．北京：电子工业出版社．1996：121～138
[2]张家兴．可持续农业中21世纪初的IPM．植物保护．1998,24(6)：38～40
[3]Pursley W.Insect electrocutor light traps.AIB Tech.Bull.1987,9:1～5
[4]Huang J X.Synthesis of the stereoisomers of the sex pheromones of the southern corn rootworm.Chemical Research in Chinese.2001, 17(1): 51～55
[5]陈庭华．斜纹夜蛾大发生原因和性诱剂的诱蛾效果．长江蔬菜．1997(7)：22～24
[6]许建．害虫成虫防治技术及应用展望．新疆农业大学学报．2000，23(S1)：32～34
[7]Robin M R.Sticky trap for monitoring leafin iners Liriomyza sativae and Liriomyza trifolii (Diptera; Agromyzidae) and their associated bymenopterous parasites in watermelon.J.of Econ.Entomol.1987,80(6): 1345～1347
[8]何忠．蝼蛄的声学通讯．生物学通报．1995,30(12)：6～7
[9]Vick, K.W..Sound detection of stored-product insects that feed inside kernels of grain.Econ.Entomol.1988,81:1489～1493
[10]Menne D.Short-time spectral analysis of the ultrasound echoes of flying insects.Localization and Orientation in Biology and Engineering,Springer-Verlag.1984:225～230
[11]沈佐锐，于新文．昆虫数学形态学研究及其应用展望．昆虫学报．1998,41(S1)：140～148
[12]于新文，沈佐锐．昆虫数字图像的分割技术研究．农业工程学报．2001，17(3)：137～141
[13]王秀丽译．昆虫学家发明害虫监听系统．世界农业．2001，11(总271)：59～63
[14]Jordan, Jay B,.Vision guided insect handling system, Modeling and Simulation.Proceedings of the Annual Pittsburgh Conference.1990, 21(5): 1995～2000
[15]Subramanyam, Bh..Alternatives to pesticides in store-prouction IPM.Kluwer Academic Publisher.2000:58～65
[16]赵志模．嗜卷书虱的试验生态研究．昆虫学报．1999,42(3)：277～282
[17]张红涛．储粮害虫图像识别中的特征抽取研究[硕士学位论文]．郑州大学．2002．
[18]张法全．基于机器视觉和小波分析的农田害虫图像识别系统[硕士学位论文]．郑州大学．2003．
[19]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究．Ⅰ．在目级阶元上的应用研究昆虫学报2003 vol46(1)45-50．
[20]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究．Ⅱ．在总科阶元上的应用研究昆虫学报2003 vol46(2)201-208．
[21]赵汗青，沈佐锐，于新文．数学形态学在昆虫分类学上的应用研究Ⅲ．在科阶元上的应用研究 昆虫学报2003 vol46(3)339—344．
[22]于思勤，孙元峰．河南农业昆虫志．北京：中国农业科技出版社．1993
[23]朱志刚，林学訚，石定机等．数字图像处理．北京：电子工业出版社．2002
[24]章毓晋．图像工程(上册)——图像处理和分析．北京：清华大学出版社．2000
[25]章毓晋．图像工程(下册)——图像理解与计算机视觉．北京：清华大学出版社．2000
[26]田村秀行[日]．计算机图像处理技术．北京：北京师范大学出版社，1986
[27]何斌数字图像处理北京：人民邮电出版社2003
[28]崔屹．数字图像处理技术与应用．北京：电子工业出版社．1997：187～200
[29]李刚，范瑞霞一种改进的图像中值滤波算法北京理工大学学报2002，(6)376～378
[30]雷晓峰，王耀南，邓曙光．在VC++中利用多线程实现图像采集与通信微型机与应用2002，2：6～8
[31]杨威，张田文，师海峰．一种用于二值图像分割的快速聚类算法计算机研究与发展 1998，35(8)：719～723
[32]高雁飞，范九伦．一种新的灰度图像阈值分割方法西安邮电学院学报．2002，7(1)：1～4
[33]石俊生，杨卫平，白凤翔，张韬，唐伟．彩色图像分割与颜色特征选择，云南师范大学学报1999,19(6)：53～56
[34]刘志敏，杨杰．基于数学形态学的图像形态滤波红外与激光工程1999，28(4)：10～15
[35]Running, S.W., Loveland, T.R., Pierce, L.L., et al.A Remote Sensing Based Vegetation Classification Logic for Global LandCover Analysis.Remote Sensing of Environment, 1995, 51: 39-48.
[36]M.Egmont-Petersen,D.de Ridder, H.Handels Image processing with neural networks-a review.Pattern Recognition 35 (2002) 2279-2301
[37]Mario K?ppena, Javier Ruiz-del-Solar.Steady-state image processing.Applied Soft Computing 1(2001) 53～62
[38]Robert Sablatnig, Martin Kampel.Model-Based Registration of Front- and Backviews of Ro- tationally Symmetric Objects.Computer Vision and Image Understanding 87 (2002), 90-103
[39]Qing Wang, Zheru Chi, Rongchun Zhao.Image Thresholding by Maximizing the Index of Non- fuzziness of the 2-D Grayscale Histogram Computer Vision and Image Understanding 85(2002) 100-116
[40]贾振堂，李生平，贺贵明等一种基于运动边缘检测的视频对象分割新算法计算机研究与发展2003(5)：684-689
[41]吕同富，刘宝军，毕秀芝图像边缘提取的简单方法及应用计算机仿真2003 20(4)：99-101
[42]张爱华，余胜生，周敬利．一种基于边缘检测的局部阈值分割算法小型微型计算机系统 2003 24(4)：661-663
[43]Changick Kim, JenqNeng Hwang.A fast and robust moving object segment in video sequences IEEE Computer Society,1999 131～134
[44]Kevin Bowyer, Christine Kranenburg, Sean Dougherty Edge Detector Evaluation Using Empirical ROC Curves Computer Vision and Image Understanding 2001 (84), 77-103
[45]谈正，刘辉，周军．旋转及伸缩不变纹理分割算法 西安交通大学学报。1997，31(1)：32～37
[46]Yamashita, Y., 0gawa, H.Relative Karhunen-Loeve transform.IEEE Tans.on Signal Processing, 1996, 44 (2): 371-378.
[47]Hua, Y.Liu, W.Generalized Karhunen-Loeve transform.IEEE Signal Processing Letters, 1998, 5(6): 141-142.
[48]江铭虎，季文铎，林碧琴等．一种特征压缩及分类神经网络的研究．电路于系统学报，1997，2(3)：18-23．
[49]Kung, S.Y.Adaptive principal component extraction(APEX) and application.IEEE Trans.Signal Processing, 1994, 42 (5): 1202-1217.
[50]Defries, R.S., Townshend, J.R.G.NDVI-derived L and Cover Classification data.Global Scale.International Journal of Re-mote Sensing, 1994, 15 (17): 3567-3586.
[51]邱道尹，张成花，张红涛等，神经网络在储粮害虫识别中的应用，农业工程学报
[52]唐向宏，贺振华，杨绍国．利用小波变换进行地震数据压缩．成都理工学院学报，1999，26(2)：183-186．
[53]张永胜，郁可．基于小波神经网络的光谱数据压缩与分类研究．计算机研究与发展，1999，36(8)：973-977．
[54]向阳，龚新高．推广模拟退火方法及其应用．物理学进展，2000，20(3)：319-334．
[55]Sidelecki, W., Slansky, J.A note on genetic algorithms for large-scale feature selection.1989,10 (5): 335-347.
[56]Roberto, B.Using mutual information for selection features in supervised neural net learning.IEEE Trans.on NN, 1994, 5 (4): 537-550.
[57]边肇祺，张学工．模式识别(第二版)．北京：清华大学出版社，2002．
[58]金连文，徐秉铮．基于多级神经网络结构的手写体汉字识别．通信学报，1997，18(5)：21-27．
[59]张学工．关于统计学习理论与支持向量机．自动化学报，2000，26(1)：32-42．
[60]Widrow B, et al.Neural networks application in industry, business and science Communication of the ACM.1994,37:93～105.
[61]高大启．有教师地线性基本函数前向三层神经网络结构研究．计算机学报．1998，21(1)80～86．
[62]李平，刘炯．前馈神经网络的结构与范化，长春光学机械学院学报．1997，20(2)：63～67
[63]阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社．2000：45～78
[64]黄德双．神经网络模式识别系统理论．北京：电子工业出版社．1996：121～138
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |