A hybrid adaptive block based compressive sensing in video for IoMT applications

The advent of advanced video communication technologies has enabled the new paradigm Internet of Multimedia Things (IoMT) wherein connected devices share multimedia contents across the internet. New approaches in compression techniques are constantly being evolved to circumvent the limitations in storing and transmitting uncompressed video. The Compressive Sensing (CS) technique that simultaneously senses and compresses an image is now considered as a better alternative for image / video compression. In Block Based Compressed Sensing (BBCS) technique, to reduce the computational complexity of CS, a large-sized image is divided into number of small non-overlapping blocks and each block is compressed at a uniform sampling rate. Instead an Adaptive Block Based Compressive Sensing (ABBCS) technique effectively make use of resources by sensing each block at a unique sampling rate adaptively estimated based on its image features. In this paper an efficient hybrid ABBCS (HABBCS) technique is proposed. HABBCS first adaptively estimates the block-wise sampling rate in a frame and the frame average is computed horizontal ellipsis The sampling rate for the entire video is then estimated from the frame averages the video is compressed by employing BBCS. The performance metrics Peak Signal to Noise Ratio (PSNR), Video Structural Similarity Index (VSSIM), Delta E and Computational Time (CT) are computed and tabulated. The results confirm that HABBCS is faster with a performance equal to ABBCS.

Automated summary

This paper introduces an efficient hybrid ABBCS (HABBCS) technique, which achieves image/video compression by adaptively estimating the sampling rate. The performance of HABBCS is comparable to ABBCS.