文档介绍：摘要
工具痕迹检测,作为一种对工具痕迹图像分析的重要方法,在产品质量检测、
安防、刑事侦查等领域有着及其广泛的应用。传统工具痕迹检测方法主要是通过专
业人员在比对显微镜下逐一地对痕迹图像进行两两比对来实现的,但是随着工具痕
迹检验对象的种类和数目的增加,受人为影响较大。效率太低,可操作性差,故这
种方法已不能满足现代高效的要求。纹理是表达物体表或结构的属性,是工具痕迹
图像的基本特征。随着计算机和人工智能技术的发展,借助纹理分析方法对工具痕
迹图像进行特征提取,从而实现工具痕迹检验标准化、自动化。
本论文主要研究一类工具加工痕迹,如平角钻头、铣刀等在生产加工过程中在
工件表面所留下的工具痕迹。针对这类工具加工痕迹的识别,建立了一种具有普适
性的研究方法,初步实现这类工具加工痕迹检验识别。首先,论文对所研究的这类
工具加工痕迹图像进行了预处理算法的比较和分析。其次,根据这类痕迹图像的纹
线特征,通过提取这类纹线来确定圆心,进而将这类工具加工痕迹转换成类直线纹
理;接着,采用频谱分析方法中的小波变换对图像进行特征提取和图像识别,设计
相应算法,计算特征参数;最后,通过 VC 平台上编程实现上述过程。该方法不仅
能够初步实现此类工具加工痕迹图像的识别,而且有效的减少了传统工具加工痕迹
识别中主观性强、耗时长、可操作性差等问题。
关键词:工具加工痕迹图像识别纹理分析特征提取小波变换
I
Abstract
Tool-marks detection, an important method to analyze the tool-marks images, has
been made full use in security-monitoring and quality general,the classical
methods applied in traces detection are the ways that pare the traces with
microscope manually. These identification rechniques are based on qualitative
analysis,which have disadvantages of gaining little information and low efficiency.
Therefore,such methods can not satisfy the need of high efficiency and effectiveness.
Texture in the tool traces image implicates much imformation of the tool and reflects the
basic features,so rexture analysis is the foundation of tool traces inspection. With the
development puter technique and artificial intelligence, realizing automatic and
shandard inspection by the feature extraction in the feature analysis measures.
The main idea of this dissertation is to apply pattern recognition and texture
analyzing algorithms to establishe the metohs for the specific tool-processing-traces,the
ones generated from flat foot drill bit,milling cutter and so on. First of all,Analysis and
Comparison of Algorithms for the image preprocess. Secondly,based on the characteristics
of these tool-processing-traces images,we extract the center of these arc-shaped tool
traces,then transform the arc-shaped tool-processing-traces into nearly-straight-l