Enhanced RGB-Based Basis Pursuit Sparsity Averaging Using Variable Density Sampling for Compressive Sensing of Eye Images

Abstract

Compressive sensing (CS) plays a critical role in sampling, transmitting, and storing the color medical image, i.e., magnetic resonance imaging, colonoscopy, wireless capsule endoscopy, and eye images. Although CS for medical images has been extensively investigated, a challenge remains in the reconstruction time of the CS. This paper considers a reconstruction of CS using sparsity averaging (SA)-based basis pursuit (BP) for RGB color space of eye image, referred to as RGB-BPSA. Next, an enhanced RGB-BPSA (E-RGB-BPSA) is proposed to reduce the reconstruction time of RGB-BPSA using a simple SA generated by the combination of Daubechies-1 and Daubechies-8 wavelet filters. In addition, variable density sampling is proposed for the measurement of E-RGB-BPSA. The performance metrics are investigated in terms of structural similarity (SSIM) index, signal-to-noise ratio (SNR), and CPU time. The simulation results show the superior E-RGB-BPSA over the existing RGB-BPSA at an image with a resolution 512 × 512 pixels into a measurement rate 10% with SSIM of 0.9, SNR of 20 dB, and CPU time of 20 seconds. The E-RGB-BPSA can be a solution to massive data transmissions and storage for the future of medical imaging.