基于偏振与颜色信息的视觉测量图像融合、评价及应用研究

The non-contact measurement method of vision measurement was widely used in automated production and manufacturing industry. The common vision measurement method was limited in application, due to the infulence of intensity and uniformity of illumination. The measurement targets are almostly compsed of metallic materials. Polarization vision is usually a beneficial supplement to normal vision measurement, and is used to discriminate the metallic and dielectric materials. Therefore the research target of this project is to develop and complete a new vision measurement technology based on the fusion of polarization and color information. "The study of image fusion, evalution and application for color and polarization based vision measurement" is this application subject. Emphasis is laid on the study of develop the vision measurement equipment based on the color and polarization information. According to the different responses of polarized and spectral reflect light, and using the property of polarization images were not influenced by the intensity and uniformity of illumination, color images are captured by the method of optimum wavelength illumination. Then the high contrast color images are obtained by the fusion of color images and Degree-of-polarization and Phase-polarization images. The study objective is to enhance the contrasts of measurement targets and backgrounds in images, to cause the fusion images close to the human being perception, to ease the subsequent tasks of image processing and understanding, to extend the application of vision measurement method and to enhance the measurement resolution.

在工业制造业和自动化生产中，大量采用了非接触式的视觉测量方法。常用的视觉测量方法受光照不均和光照过暗的影响，视觉测量的适用范围受到限制。视觉测量的对象大都是以金属材料为主，而光的偏振对于金属和绝缘体材料的分辨极为明显,是普通视觉测量的有益补充。因此本项目研究的目标是发展和完善一种以偏振与颜色信息结合的新视觉测量技术。本申请课题～"基于偏振和颜色信息的视觉测量图像融合、评价及应用研究"，研究基于偏振与颜色信息的视觉测量装置，根据测量目标材料对于光的偏振和光谱反射的不同反应，利用偏振图像不受光照不均匀与过暗的影响，采用最优波段照明的方法获得高对比度的彩色图像，将获得的彩色图像与偏振度、偏振相位图像融合，得到真彩色的高对比图像，增大图像中测量目标和背景的差异，提高图像的对比度，使得图像更接近于人的视觉感知习惯，减轻后续的图像处理和图像理解任务，增加视觉测量的适用范围，提高测量时的分辨力。

在工业制造业和自动化生产中，大量采用了非接触式的视觉测量方法。常用的视觉测量方法受光照不均和光照过暗的影响，视觉测量的适用范围受到限制。视觉测量的对象大都是以金属材料为主，而光的偏振对于金属和绝缘体材料的分辨极为明显,是普通视觉测量的有益补充。本项目发展和完善一种以偏振与颜色信息结合的新视觉测量技术。本课题～"基于偏振和颜色信息的视觉测量图像融合、评价及应用研究"，研究基于偏振与颜色信息的视觉测量装置。首先为了提高被测目标物体信息获取的有效性，对光源进行研究设计，提出了一种基于LED阵列与漫反射自由曲面的照明光源设计方法，使得目标平面具有较好的照度均匀性以及较高的光能利用率，并且获得了一系列的研究成果；其次在获取被测目标物体图像信息的基础上，以目标物体在不同光源下采集的图像呈现出不同的颜色值，从而导致图像与目标物体出现视觉上的偏差为研究对象，提出了一种图像矫正与图像特征提取算法，使得矫正后的图像更接近于目标图像，并对系统进行了校正；最后选取了相关实验对象，探索了在非接触视觉测量领域，基于颜色与偏振的视觉测量方法，实验结果显示图像的对比度提高，减轻后续的图像处理和图像理解任务，增加了视觉测量的适用范围，提高了测量时的分辨力。