人体光学测试中用于消除测量位置改变的掌/指纹图像匹配技术(英文)

Correction of Measured Position Changes in Optical Probe-Based Measurement Using Palm/Finger Print Image Matching Technique

应用光学测头的测量方式中,由于方便测量和易于随时测量,手掌和手指经常被选作测量部位.然而,由于人体不同部位皮肤结构的变异性,由人为运动导致的测量位置改变会对被测光有很大影响.为此,开发了一种基于掌/指纹图像匹配技术的有效方法,以避免测量位置的改变.该方法通过应用小波变换、相位相关和曲面拟合技术,分析不同测量位置之间旋转和平移改变量,然后驱动光学测头到达相应的位置,从而使光学测头每次均测量同一位置.其精度评测结果表明,平移检测精度达到1/50像素,旋转检测精度达到0.15°.光谱测量结果表明,该技术的应用使光谱稳定性提高了3倍左右.

Palm and finger are usually preferred to be used as measured sites in optical probe-based measurement because they both are convenient and easy to be measured at any time. However, measured position changes induced by artificial movement could influence the detected light due to the inhomogeneity of skin tissue structures beneath different measured sites. In this paper, our aim is to present an efficient method to correct measured position changes by using palm/finger print image matching technique. This image matching technique combined wavelet transform with phase correlation and paraboloid surface fit. During the measurement, optical probe was driven to the fixed position based on the translation and rotation changes between different measured positions calculated by the proposed technique. The results show that for the noise-free palm/finger print images, the matching accuracy is less than 1/50 pixel for translation and 0.15° for rotation. Finally, spectroscopic measurement results show that spectral stability with current palm/finger print image matching technique increases by around three times more than that without the technique.