In recent years, digital image processing technology has been widely used in the textile industry. In order to achieve highprecision automatic detection of the fabric warp and weft density, this paper designed a set of miniaturized embedded fabric weft density detector. The detector uses ARM architecture processor as the core hardware platform and running on the Linux operating system. The detector uses microoptical lens to capture fabric images, and uses histogram equalization, frequency domain filtering and other image processing algorithms to detect the fabric warp and weft density. In this paper, the hardware design part introduces the system framework and hardware structure of the detection system in detail. The software design part introduces the driver design scheme of the high speed CCD image sensor based on the FIQ interruption system, and the weft density detection algorithm which synthesized various image processing algorithms. Experiments show that for the twill weave denim, the relative error between manual measurement and automatic measurement is less than 1%. The test results are satisfactory and the detector can be mass produced.