A real-time object detecting system using compressive sensing

Nowadays, Wireless visual sensor Network (WVSN) is gaining significant importance and has several distinctive applications. WVSNs provides better results when it come to object detection since it consists of several tiny camera sensors which can capture the object's image and helps in identifying the object more precisely. There are some limitations of WVSN like battery constraints, limited storage memory so to overcome these limitations the proposed method is using block based compressive sensing technique (BCS) which applies compression only to the required blocks of image rather than applying it on whole image frame. The proposed method also incorporates a change detecting algorithm which will work only if the Mean square error (MSE) comparator crosses its predefined threshold value. The main aim of the proposed method is to achieve faster reconstruction of image with acceptable image quality so, to attain the required performance contourlet transform and projection driven compressive sensing recovery technique has been used. Also to increase the speed of reconstruction wiener filter is incorporated into iterative projected landweber compressive sensing recovery technique. The results have shown that the proposed block based compressive sensing with smooth projected landweber performs better than other methods as compared to quality of the reconstructed image.