The research of image recognition's high-speed processing problems is conducted to improve the image's practical processing capabilities. Most of the existing image recognition system is composed on the basis of ordinary CPU machine or DSP, which has the defects of limited calculation bandwidth, low efficiency, resource wasting and poor flexibility. The reason is that, due to the commonly used computer hardware design's lack in image recognition algorithm, image processing algorithm's data is low in number and has Bit Boolean operators, convolution and so on. All of these make it inappropriate...
