Optimal feature vector can be extracted in color image space. To acquire optimal parameters of pulse coupled neural network (PCNN) for color image space, color image processing are carried out by simulating visual perception of mammal. Entropy sequence of PCNN is calculated as feature vector to classify the traffic signs in RGB and HSV color space. Optimal parameters of PCNN are acquired through experiments. Experiments are carried out in GB5768-1999 of standard traffic signs image database. Experimental results show that the maximum inter-class distance value is acquired among 43 warning signs, 42 prohibition signs, and 29 instruction signs based on the entropy sequence of PCNN in blue color space while the optimal parameters of PCNN are αL=1, αF=0.1, αE=1, VL=0.2, VF=0.5, VE=27, β=0.1, and N=50 iterations. Blue component of the RGB model can fully reflect the color characteristics of traffic signs, entropy sequence of PCNN can be used as the vector sequences to distinguish three categories, superior to conventional color image into a grayscale image processing methods.