A beef image on-line acquisition system based on TMS320DM642 was developed according to the technological requirements of beef automatic grading. The rib-eye beef images were collected online by using the developed system. Then a mask was constructed for background segmentation of each acquired image. As DM642 adopted three chrominance components of YCbCr model to display and save images respectively, in order to reduce calculating time, only Cr chrominance component of beef image was considered in the following image processing. After the Cr chrominance component of beef images were binarized, region labeling algorithms and small regions removing method were employed to extract the effective rib-eye region. Experiments showed that the proposed DM642 system could be used for beef image acquisition online and rib-eye region segmentation. The average extraction accuracy of effective rib-eye region was 92.9%.