The invention discloses a 
machine vision-based fabric 
physical property detection method and device. The 
machine vision-based fabric 
physical property detection method comprises the following steps: a, arranging an area-array camera with a 
telecentric lens and an LED (light-emitting 
diode) 
light source in a face-to-face manner, and forming a detection area between the area-array camera and the LED 
light source; b, controlling the area-array camera by a computer; c, enabling a fabric to be detected to pass through the detection area and the LED 
light source to irradiate the fabric to be detected, and photographing the fabric to be detected by the area-array camera; and d, transmitting images photographed by the area-array camera to the computer, converting the images of the fabric to be detected from 
time domain images to 
frequency domain images by the computer, extracting peak position coordinates of the horizontal direction and / or vertical direction of center position coordinates from the 
frequency domain images, calculating the distance between the peak position coordinates and the center position coordinates, and then calculating the density value of the fabric to be detected and / or the weft bevel angle of the fabric to be detected. The 
machine vision-based fabric 
physical property detection device comprises a framework, an LED light source and an area-array camera, wherein a 
telecentric lens is mounted on the area-array camera. According to the 
machine vision-based fabric physical property detection method and device, the images of a fabric flowing on the 
production line can be acquired, and high detection accuracy is achieved.