【作者】 尚可可；

【导师】 刘迎；

【作者基本信息】 天津大学 ， 光学工程， 2007， 博士

【摘要】 皮肤红斑(Erythema)是皮肤病常见的临床症状,对皮肤红斑颜色和形状的定量测量可以为一些皮肤病病变程度和用药效果提供客观的评价方法。本论文选取皮肤红斑作为研究目标,研究它的客观化测量方法。我们的主要工作有:一、设计一套皮肤病彩色数字成像测量系统。通过对比HSV色彩空间、RGB色彩空间、Lab色彩空间的颜色参量,发现HSV色彩空间更适合进行皮肤颜色测量。在分析大量皮肤图像样本之后,为了使HSV色彩空间更适合皮肤颜色描述,对色度分量进行修正。用计算机模拟的方法分析了测量系统的精度,模拟结果显示测量精度和被测对象的颜色有关。二、在Dawson皮肤光学模型的基础上,采用正常部位与红斑部位光强倒数的对数差作为参数来描述皮肤红斑变化。理论推导表明:该参数与正常部位和红斑部位真皮层中的血色素浓度差成正比,可以较好地反映红斑的病变情况。然后对该参数的光强稳定性实验结果进行单因子方差分析,结果表明该参数能消除入射光强变化的影响。最后运用计算机模拟的方法对该参数的测量精度进行了讨论。三、将模糊C均值算法(FCM)用于皮肤红斑图像分割,选择了HSV色彩空间作为分割的特征空间。为了抑制噪声对分割结果的影响对该算法进行了改进,实验表明改进后的算法可以得到更好的分割效果。讨论了皮肤红斑图像分割时的降维方法,提出了皮肤红斑图像分割的一维分割特征新参量。对新参量与一维分割特征参量h、s和由h、s构成的两维分割特征参量的分割效果进行比较,实验结果显示,采用分割特征新参量可以得到更好的分割效果。更多还原

【Abstract】 Skin erythema is a familiar symptom in clinical dermatology. Its objective measurement is of importance in clinical dermatological diagnosis and new drug tests. This paper studies the ways to measure the skin erythema objectively.Since skin erythema covers the information of color and shape, we adopted the digital imaging way to measure it. We studied the precision of the measurement system, the measurement indexes, the way how to measure and the methods to segment the color skin erythema images.Firstly, we introduced an improved digital imaging measurement system on skin diseases and compared the HSV, Lab, RGB color space to pursue which color space is fit to measure the skin color. The result demonstrates that the HSV color space is of superiority. Then, we’ve corrected the HSV color space in order to suitably describe the color characteristics of skin erythema. At last, we analysed the precision of the measurement system through the computer simulation.The result suggest that there is certain connection between the precision of the measurement system and the color of the measured object.Secondly, we investigated a new parameter defined by the difference of the logarithm of the reflected light intensity between normal skin and the erythema skin to describe the pathology change of the erythema. Deducing theoretically, we can draw the conclusion that this parameter not only can reflect the deviation degree of the concentration of haemoglobin in the dermis from the normal value, but also can hardly be influenced by the melanin in the epidermis and the incident intensity. In the aspect of experimental study, we first analyzed whether this parameter can be affected by the incident intensity, and then we applied it to the clinic and traced the effect of curing skin erythema.Thirdly, we adopted the FCM arithmetic to segment the color skin erythema images and e proposed K (k=s/h, h is hue, s is saturation of HSV color space) as the segmentation feature parameter. By comparing the effect and the speed of the segmentation when the k and other three segmentations feature were employed, we found that the result by choosing K as the segmentation feature parameter is better.更多还原