Abstract: In this paper, a hardware-accelerated volume rendering method on standard PC platform is realized, to visualize the Chinese Digital Human datasets and American Visible Human datasets. By the method, during the preprocess stage of Chinese Digital Human, in order to remove image background, firstly the image is converted from RGB color space to YC_bC_r color space, then the Cb channel data is handled with Otsu method to generate binarized images. At last the image background would be removed through using the previous output results as masks. In order to render volumes in arbitrary size, a block-algorithm is performed, and the blocks are constructed by BSP Tree to get a proper rendering order. When designing transfer function, we convert the images into CIELUV color space and compute the GVF magnitude of the L channel data as the transparency. Experiments show that the results of the algorithm are satisfactory and efficient.