## how the compression ratio is calculated

on 24 Dec 2012

### Wayne King (view profile)

```i have a code
```
```X=imread(file);
```
```wavelet='haar'
```
```level=2;
```
```[c,s] = wavedec2(X,2,wavelet);
```
```alpha = 1.5; m = 2.7*prod(s(1,:));
```
```[thr,nkeep] = wdcbm2(c,s,alpha,m)
```
```[xd,cxd,sxd,perf0,perfl2] = wdencmp('lvd',c,s,wname,n,thr,'h');
```
```disp('Compression Ratio');disp(perf0);
```
```the size of xd and X is same,plz tell how compression ratio is calculated
```
```both images look like same ,how we can say it is compressed ny 84%
```
```perf0 is 84%
```

## Products

No products are associated with this question.

### Wayne King (view profile)

on 24 Dec 2012

perf0 is the percentage of zeros in the wavelet coefficients. So 84% indicates that 84 percent of the wavelet coefficients have been set to 0.

perfl2 is the ratio (as a percentage) between the L2 norm of the compressed wavelet coefficients and the original (before compression) wavelet coefficients.

Yes, the output compressed image is of course the same size as the input, the point is that it is constructed from a much smaller set of wavelet coefficients

FIR

### FIR (view profile)

on 24 Dec 2012

nnz(perf0) gives as 1

Wayne i sthere any compression that can make the image into smaller size,the code used me as of 3 to 4 lines as above,

Wayne King

### Wayne King (view profile)

on 24 Dec 2012

perf0 is just a scalar, so why would you use nnz() on that and expect something other than a 1?

FIR

### FIR (view profile)

on 24 Dec 2012

Wayne i sthere any compression that can make the image into smaller size, as of 3 to 4 lines as above,i tried huffman but matlab crashes

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi