一种基于光学干涉原理的图像分割方法

Measurement of image segmentation based on optical interference

介绍了基于光学干涉原理的图像分割理论算法。在计算机上模拟一个均匀相干光源、一个滤波器和一个接收屏。选择强度 相位的转换曲线,将输入图像的强度信息转化成伪相位信息,并显示在滤波器上。利用光源照明滤波器上的相位图,在接收屏上得到其干涉条纹。再通过选择阈值对干涉图进行分割得到二值图,从而实现输入图像的边缘分割。计算结果表明,该方法得到的图像轮廓,与Sobel算子效果相当。与实验结果对比分析可得,改善硬件并实现良好的分割效果的关键在于:控制光源的相干长度,或增大液晶的点阵间隔,并利用CCD相机和计算机相连,对干涉图进行二值化。

A kind of measurement of image segmentation based on the principle of optical interference is presented and realized by simulation in computer. A uniform coherent light source, a filter and a receiving screen are designed in computer. The pseud phase distribution, which is translated from intensity distribution of the image by a conversion rule, is shown on this given filter. The interference pattern is produced when the filter is illuminated with the light source, and sampled on a receiving screen. Choosing a proper threshold segments the interference pattern and the edge of the input image can be obtained. Numerical simulation shows the effect of this segmentation method is equivalent to that of Sobel operator.Contrast between simulated results and experimental results is analyzed. It indicates two ways to improve the segmentation effect of the hardware system, one is to control coherent length of the light source, or to enlarge lattice distance of the liquid crystal screen; the second way is that CCD camera be connected with a computer to change the interference pattern into binary image.