不同转换方式对三角函数算子分割效果的影响

Influence of intensity-phase conversions on image segmentation by trigonometric function operator

提出了一种新的图像分割算法三角函数算子,它是一种基于光学干涉原理的快速图像分割方法,并分析了其硬件实现装置。在该算子的运用中需要将输入图像的强度信息转化为相应的相位信息。定义了不同的强度-相位转换方式,例如线性方式、对数函数方式、正切函数方式、反正切函数方式等。通过数值计算,研究了不同转化方式对三角函数算子分割效果的影响。结果显示,不同的转换方式及其参数,都直接影响该算子的分割效果和边缘类型分辨能力。分析表明,在对数S形函数方式下,不但能够检测阶梯状和脉冲状等类型的边缘,还能够检测出屋顶状边缘。

The Trigonometric-function operator is put forward as a novoel image segmentation arithmetic. It is deduced by the principle of optical interference and applied in fast image segmentation. The hardware of Trigonometric-function operator is analyzed. The pseud phase of input image is important for the operator and need to be translated from the intensity distribution by appropriate conversion rule. Various conversion manners are defined, including linearity, logarithm function, tangent function and are tangent function. Numerical simulation results show that image segmentation effects of trigonometric-function operator vary with the manners of intensity-pseud phase translation and their parameters. When logarithm sigmoid tranfer function is used as in- tensity-phase conversion manner, the trigonometric-function operator can not only detect edge style of ladder and impulse, but also detect edge styles of housetop.