To detect the diameter and number of tiny holes in chemical fiber spinneret,a method based on machine vision was proposed and verified.Firstly,the realtime image of tiny holes in chemical fiber spinneret was acquired by highresolution industrial camera.After grayscale transformation,the image edge was enhanced by GaussLaplace operator and then extracted by Canny operator.Next,the edge was fitted by circle to judge whether there are burrs or inadequate roundness in the tiny holes.Finally,based on this method,a detection system for tiny holes in chemical fiber spinneret was developed,which included image acquisition device,spinneret transmission device and detection software.Test results of the system show that this method can be used to continuously detect the diameters and number of the tiny holes in spinneret without contact,and is accurate at micronlevel.