An Adaptive-ADMM Algorithm With Support and Signal Value Detection for Compressed Sensing

Optimal-correlation-based reconstruction for distributed compressed video sensing

“…Pudlewski et al briefly discussed challenges involved in the transmission of video over a WMSN [15] and presented a cross-layer system that jointly controls the video encoding rate, the transmission rate, and the channel coding rate to maximize the received video quality [16]. Besides, a dictionary generation scheme for CS-based video sampling and a dictionary learning based DCVS reconstruction method were proposed in our previous work [17,18] respectively, and more recently, an adaptive alternating direction method of multipliers with its application to compressed video sensing was presented in [19,20].…”

“…(25) can be easily solved by employing a direct method [19]: (i) based on the Cholesky factorization, we have lU T 2 U 2 þ kI ¼ L Ã L T with a lower triangular matrix L; and (ii) f kþ1 is computed as…”

Joint Sampling Rate and Bit-Depth Optimization in Compressive Video Sampling

“…Although ADMM was originally proposed in the mid-1970s [20], it has been widely known until recently with the development of large-scale distributed computing systems and massive optimization problems, and is well suited to the large-scale reconstruction problem arising in CS. For example, an adaptive ADMM reconstruction algorithm was proposed in our previous work [21] for compressed video sensing with support and signal value detection.…”

“…9(a). In the frame-level rate allocation method, the bits used for the reference frame is computed based on the target measurement bit-rate via a simple model (21) and the remaining bits are allocated equally to the non-reference frames. In this way, given the target rate , and can be firstly estimated for the reference frame.…”

“…In our algorithm, the sampling ratio and bit-depth of the reference frames are estimated via (21), while the values of and for non-reference frames are determined by using the proposed RDO based rate allocation algorithm, wherein the empirical values of model coefficients in (21) are set to , ,…”