Abstract:In the course of using standard metal tank for liquid metering, the traditional method of measuring the level of human-readable is inefficiency and inaccurate.Putting forward a high-precision standard metal tank level image recognition device which is mainly composed of the level adjustment and image processing,choosing S3C2440 ARM chips and Linux Embedded system as a hardware and software platform. level adjustment is implemented by measuring the inclination of tank by SCA100T axis inclinometer sensor and controlling stepper motor to automatically adjust, realizing closed-loop control; Processing the acquired liquid level image by OpenCV,including graying, filtering, edge detection, liquid extraction, and then after converting into actual level height through calibration. Experimental results show that this method can measure the level with high speed and precision, and it have a better promotional value used in fuel dispensers verification device.