【作者】 李峰泉；

【导师】 曾衍钧；

【作者基本信息】 北京工业大学 ， 流体力学， 2001， 硕士

【摘要】 本课题主要是应用小波变换来解决基于掌纹图象的癌症预测系统的建设中的图象处理问题。众所周知，癌症是一种遗传疾病。而人的掌纹纹理在一生中都是不变的。有理论提出个体的掌纹的纹理可以在某些方面表现出他的部分基因信息。因此，有假设说可以通过研究个体的掌纹纹理特征来早期诊断癌症。本课题就是建立在这个假设之上，希望建立掌纹纹理类型的一个分析系统，用来向医生提供掌纹诊断的依据，并希望最终找到一些掌纹的特征能够和癌症基因相联系。 所谓的OLAP系统，就是能够既存储数据也能够对所存储的数据（或称作信息）进行实时分析处理的数据库管理系统DBMS（DatabaseManagement System）。构建一个OLAP系统的主要目的就是随时向客户（具体到这里是医生）提供数据的汇总和分析报告，用来支持客户的决策。 整个系统是课题组共同来开发，我主要在工作组中负责开发一套由6个FIR滤波器组成的滤波器组。这套滤波器用于三通道小波变换的分解滤波器和重构滤波器。在整个系统中，它们主要用在两个方面：预处理部分对采集来的图象进行几何校正和消噪，在OLAP系统向客户提交报告时形成分析结果的缩略图。 具体到滤波器组的设计，我提出采用优化方法来设计分解滤波器组，通过设计符合适当条件（分频点、截止频率等）的低通半带QMF滤波器，然后通过合适的仿酉变换求出整个滤波器组。更多还原

【Abstract】 a system of palm-print image online analysis managed to be established for cancer prediction. It was well known that the cancer is an inherited disease. The palm-print of one person will not change its textile during this people抯 whole life. A theory said that the gene of one individual is expressed by palm-print textile in some way. The assumption of this project is based on that there has already been stable relationship between cancer-gene and palm-print. The arm of this system is to analysis the textile of palm-print to find if there抯 any association between some kind of palm-print pattern and cancer gene. An OLAP system is a kind of DBMS (Database Management System), which can perform the function of both data (information) storage and data analysis. To establish an OLAP system is to provide client (as to this system is doctors) the summary and analysis of data stored to support the decision procedure. The whole system is a joint work. So everyone in the workgroup can only do a part of system development. The part I do is to develop a wavelet filter banks, which Consist of 6 FIR filters. These filters are combined as two sets of three-channel wavelet transform filters that are used for wavelet decomposition and wavelet reconstruction. The three channel wavelet filters are used in two part of OLAP system: first, the pre-processing stage of image, to geometry rectification the image captured and de-noise it as well as. Second, the summary committing stage of OLAP system, to make the thumbnail of the image analysis result. As to the implementation, I provide an idea that to use optimize method to develop the qualified (frequency-divide point, stop-frequency, and so on) half- band low-pass QMF filter. Then make out all of the filter banks via suitable pseudo-unity transform.更多还原