Parameterizations that use mesh simplification to build the base domain always adopt the vertex removal scheme.This paper applies edge collapse to constructing the base domain instead.After inducing the parameterization of the original mesh over the base domain,new algorithms map the new vertices in the simplified mesh back to the original one according to the edge transition sequence to integrate the parameterization.We present a direct way,namely edge classification,to deduce the sequence.Experimental results show that the new parameterization features considerable saving in computing complexity and maintains smoothness.
This paper gives a practical schema for using DSP boards to construct Vehicle License Plate Recognition (VLPR) modules that could be embedded in any Intelligent Transportation System (ITS). Using DSP can avoid the heavy investment in dedicated VLPR system and improve the computational power compared to PC software environment. Low cost, high computational power, and high flexibility of DSP provide the License Plate Recognition System (LPRS) an excellent cost-effective solution to execute the major part of the recognition tasks. This paper describes a successful implementation of VLPR system based on Texas Instruments (TI)'s TMS320DM642. The DSP board acquires video (which could be output to a monitor for surveillance) from a camera, captures images from the video, locates and recognizes the license plates in images, and then sends the recognized results and related images after compression to a host PC through the network. Finally, the overall software is optimized according to the features of DM642 chip. Experiments showed that the DSP VLPR system performs well on the local license plates, and that the processing speed and accuracy can meet the requirement of practical applications.
