【作者】 丁剑；

【导师】 彭小奇；

【作者基本信息】 中南大学 ， 物理电子学， 2005， 硕士

【摘要】 小波分析因其独特的时(空)-频多尺度分析能力,在信息处理领域得到了广泛的应用与发展。图像处理作为信息处理的重要组成部分,成为小波分析的重要应用领域。本文以高温熔体表面温度场的测试为研究背景,综合运用CCD技术、非接触式测量技术,结合小波变换,对高温熔体在线测温系统进行了研究和开发。 本文结合热辐射理论、CCD器件工作原理、光学系统成像原理和计算机图像处理技术,研究了基于CCD图像传感器的测温系统的设计方法,分析了辐射式测温系统中的CCD摄像器件必须具备的性能,以及选择CCD摄像器件和视频采集卡时应考虑的主要因素。论述了基于CCD的非接触式测温技术,研究了基于CCD的比色测温原理和小波变换的图像分割技术。为将高温目标图像与其背景图像进行准确分割,提出了高温熔体的一维小波在线分割算法和二维小波在线分割算法。一维小波在线分割算法采用了图像灰度直方图小波变换取最优阈值法,得到了较好的分割效果。本文提出的二维小波分割算法通过构造出一类Bubble函数,并充分利用Bubble小波本身所具有的信号自适应能力,根据计算出的小波变换模中取阈值,可以直接快速准确地分割出目标图像。与传统的最大方差取阈值法相比,本文提出的Bubble小波分割算法具有更快、更准确的优点,与一维小波分割算法相比,对图像的细节部分刻画更准确。设计并构建了CCD测温系统,完成了对系统的标定。更多还原

【Abstract】 Wavelet analysis is obtained broad application and development in signal process, because it has especial time-frequency analysis capability. Image signal process is the important part of signal process, also wavelet analysis is applied in image signal proess. In order to measure the surface temperature field of high temperature smelting metal, the measure system is researched and developed based on CCD technique, contactless temperature technology and wavelet transform.In this paper, combined with thermal radiation theory, CCD working principle, optical system imaging theory and computer multimedia technology, the design method of the temperature measurement system is studied, and the necessary capability of CCD image sensor in this thermal radiation measurement system and the main factors in selecting CCD image sensor parts and image collection boards are analyzed as well.The theory of temperature measurement by using colorized CCD is summarized. The colormetry temperature measurement method of CCD and image segmentation technology by wavelet transform are researched, and the algorithm is designed, which can extract the target in fast. The 1-D wavelet transform of histogram of the intensity image got the best threshold, the effect of segmentation is all right. In 2-D wavelet algorithm, the bubble function is constructed,which had the capacity of signal auto-matched,and the threshold from wavelet modulus is given out. The 2-D wavelet algorithm can segment the aim image in fast, the effect is good. In this paper,compared the 1-D wavelet algorithm,2-D wavelet algorithm and the automatic threshold-taking method of the max square error, the results is show that 2-D wavelet algorithm is faster and more exact than the automatic threshold-taking method, and accurater than the 1-D wavelet algorithm.The CCD measurement system is calibrated.更多还原