Information Technology Reference
In-Depth Information
Fig. 12.23
(
a
) Lena image used in image compression tests (
b
) Reconstructed Lena image
achieved by a one hidden layer FNN with sigmoidal basis functions
Fig. 12.24
(
a
) Reconstructed Lena image achieved by an FNN with Hermite basis functions,
(
b
) Boats image used in image compression tests
put together to form a training matrix X of size n P . The target matrix D for the
training of the neural net is considered to be the same as the input matrix X.The
structure of an FNN for image compression is depicted in Fig.
12.22
b. To quantify
the quality of the reconstructed image, the compression ratio can be used, which
in a simple form is expressed as:
L L
m P C m n
D
(12.28)
