In order to realize the detection of mango size, maturity and degree of decay, a test platform based on DSP6437 development board was constructed. Image acquisition was carried out on this platform, with conversion and collection of the data stream. Therefor, the RGB and grayscale images were obtained, the super neighborhood average method for smoothing the image was used to detect the region of interest, and according to the regional average gray degree of maturity, the classification of mango was determined. The edge points of mango were calculated by Laplace transform, and the minimum envelope rectangle was rotated to judge the size. Combining the visual inspection with gray histogram statistics and the taste sensor, the ripeness of mango was determined. The experimental results showed that the detection platform was small, stable, accurate and more suitable for actual production detection, and was also practicable for the inspection of mango production.