Abstract: Compared with traditional PC-type machine vision systems, embedded machine vision systems have the characteristics of low power consumption, small space, simple operation, easy expansion, and low cost.Therefore, embedded machine vision systems are widely used in the image acquisition and processing aspects.In this paper, ARM Cortex A9 is used as the hardware platform, combined with the embedded Linux system, the image is collected through the USB camera and the image processing algorithm is deployed, and the interactive software system is designed to realize the grading detection of corn seedlings.The results show that using this algorithm can effectively extract corn seedlings, and the segmentation accuracy reaches 93.87%.The classification results are displayed correctly, the running time is at about 240 ms stably, the system performance is stable, and it has broad application value.