Aiming at the problems of high power consumption and large volume in traditional PC-based binocular ranging systems, an embedded binocular ranging scheme based on the Zynq 7020 heterogeneous platform is proposed. In this scheme, the Vivado HLS tool is used to implement hardware implementation of image preprocessing algorithms. Hardware acceleration for RGB to YCbCr color space conversion, threshold segmentation, erosion filtering, and target bounding box extraction is realized through FPGA parallel computing. The PS part configures algorithm parameters and reads target coordinates via the AXI Lite interface, and completes distance calculation combined with the parallax-distance calibration formula. Experimental results show that the system error is 2%~4% at a distance of 50 cm, and the image processing speed is improved by about 15 times compared with the pure software implementation. Meanwhile, it meets the embedded application requirements of low power consumption and small size.