采用人工流体神经网络实现图象型微粒的检测

Micro-particle image processing based on an artificial neuron network with fluid properties

在二值微粒图象中，微粒的形态是一封闭的连通区域，微粒检测其实就是连通域的检测。据此，文章提出了采用模拟实际视觉系统扩散行为的人工流体神经网络来实现微粒检测的新方法，并进行了计算机模拟仿真实验，结果证明该方法检测能力强、检测结果准确可靠。

The form of macro-particle is a close successive region in two-valued micro-particle image, and thedetection of microparticles is actually that of close successive regions. Accordign to this theory, this paper presenta new technique for micro-particles detecting based on an artificial neuron network modeled on the diffusing behaviour of real visual system, which has fluid properties. The results of simulation on computer shew that thismethod is very effect.