Introduction:
Compression perception, as the name implies, is the perception of compression, which contains two layers of meaning, 1, perception, that is, acquisition or sampling, in the traditional signal acquisition, in order to not distort, must meet the Nyquist sampling theorem, in the previous blog post has introduced the compression sensing in the sampling and traditional signal acquisition of the connection and difference, referring to the HTTP ://www.cnblogs.com/andyjee/p/5050321.html; 2, compression, that is, data compression, here is a concept of difference and traditional compression, so this article mainly from the point of view of compression, the compression perception and traditional compression of the connection and difference.
Traditional compression:
The traditional signal acquisition and processing process includes: sampling, compressing, transmitting, extracting four parts, the sampling process must follow the Nyquist sampling theorem, this way sampling data is large, the first sampling after compression, wasting a lot of sensor element, time and storage space.
Compression Awareness:
The theory of compression perception is for the sparse representation of signals, which can combine data acquisition and compression.
Compression perception vs. traditional compression:
From the process point of view, the traditional compression is the first sample compression, and compression sensing is the edge sampling edge compression, that is, the sampled value is the compressed value, the advantage is that it can reduce the number of sensor elements, reduce the acquisition, transmission time, reduce storage space and so on.
From the compression (coding) way, the traditional compression is mainly through the transformation of the way (the sparsity of the transformation domain) to achieve compression (lossy compression), known as DCT transform, wavelet transform, etc., and the compression perception is based on the sparse transformation of the use of random linear projection measurement to compress (to meet the measurement matrix and sparse basis of the uncorrelated).
From the view of Reconstruction (decoding), traditional compression is mainly reconstructed by inverse transformation, and the compression perception is to recover the signal by the Nonlinear reconstruction Algorithm (optimization method).
In terms of complexity, traditional compression is difficult to compress (encode), while refactoring (decoding) is easy, while compression perception is the opposite, compression (coding) is easy, and refactoring (decoding) is difficult, so the feature is suitable for some harsh environments.
A brief talk on compression perception (13): Compression perception and traditional compression