Moving Vehicle Recognition Based On Image In Wireless Sensor Network

With the continuous development of wireless sensor networks, information-richimage data become a research hotspot. In view of battle field surveillance networkapplication background, the research content of this paper is the moving vehicledetection based on image in wireless sensor networks.For the case that camera is stationary, the paper focuses on three different methodsapplicable to this situation, the three difference method performes well performance inmoving vehicle detection combined with Ostu threshold segmentation, morphologicalprocessing and connectivity analysis. At the same time, the speed of morphologicalprocessing has been improved through the decomposition of the structural elements andthe shift operators.At the same time,the paper also studies four local invariant feature extractionalgorithms including SIFT,SURF,ASIFT and S-ASIFT. Simulation results show that thefour algorithms are fully scale invariant and rotation invariant.And compared with SIFTand SURF,ASIFT and S-ASIFT algorithms maintain good robustness with hugeperspective change.The SURF and S-ASIFT algorithms use integral images and boxfiltering to replace the original scale pyramid establishment, thus significantlyimproving their computing speed.Finally, the paper gives the overall design of the system.Then the moving vehicledetection and Jpeg compression are complished on OMAP-l38while S-ASIFT featureextraction and vehicle classification based on feature points matching are completed onthe PC side.