Image Reconstruction Using Compressive Sensing Technique for Hardware Implementation

Abstract

ADC (Analog to Digital Converter) which follows the Nyquist rate has changed signal processing. Most of the real time applications required too many samples if Nyquist rate is followed. This may involve more cost or even practically not feasible to build systems capable of acquiring samples at Nyquist rate. Compressive Sensing (CS) is a recent trend emerged as a better concept than Nyquist technique by enabling reconstruction of sparse signals which are acquired bellow Nyquist rate. The authors propose reconstruction of Image using Compressive Sensing Technique. Various matrices like Partial Hadmard, Bernoulli Matrix, Uniform Spherical and Random Matrix with proper threshold are used. The reconstruction time, SNR and MSE are measured. Experiments are carried on various sized image with Foreward Hadmard Transform. The experimental results show that for an image of size 256×256 the reconstruction time is 9 sec with signal to noise ratio 23 dB. For inage of size 512×512 the reconstruction time is 15 sec with signal to noise ratio as 26dB. This gives the opportunity to build CS hardware as an alternative for ADC.