For reducing a fixed threshold binarization method’s drawback and the serious Inadequate and Interference after binarization handling,meanwhile the traditional DSP and other processors in highspeed image processing in realtime poor is not good enough, proposing that using FPGA / Nios II processor and the improved adaptive threshold Sobel algorithm for image binarization, morphological processing again. The whole system implementation includes the original image capture, color format conversion, gradation processing, edge detection and median filtering using the weighted average method and adaptive threshold image binarization, and finally by using morphological methods to further deal with the binary image, to achieve a more efficient extraction of contour, and through VGA interface display. Experimental results show that the system is high efficient, fast, stable and reliable.