For the demands of real-time processing and application environment, a new animal vision-based vehicle identification algorithm is put forward, and packaged as component used in heterogeneous processing platform. By determining relevance of neighboring pixels, this algorithm eliminates false errors and raises the accuracy of vehicle identification. Meanwhile speediness algorithm adapting contributes to rapid application. Experimental results show that the heterogeneous processing platform reduces development period, improves algorithm utilization, fast updates algorithm and processing.