Course: CIS451/651 Data Compression
in Multimedia
Professor: Paul D. Amer
Semester: Spring 2009
Title: Homework - Chapter 13 - Transform Encoding
Due Date:
Tasks
Read Chapter 13. Omit Section 13.4.4.
In solving these problems, you may write your own programs, or use any
public software packages such as matlab. I recommend the latter!
-
(4 pts) Consider Sayood's Example 13.2.1.
-
(submit) Repeat this example with an angle of 30 degrees.
-
(submit) Compute the mean squared error introduced in the reconstructed values using
30 degrees, and compare it with the error in the book's example that used
approx. 68 degrees (arc tan 2.5).
-
(2 pts) The book says the transformed values in Table 13.2 are "energy
preserving".
- (submit) Explain "energy-preserving".
- (submit) Verify Table 13.2 is energy-preserving.
-
(3 pts) Chapter 13: Projects and Problems: 2
-
(2 pts) Compute all coefficients of the 2X2 Discrete Cosine Transform (DCT)
transform matrix
-
See equation 13.43
-
Solve completely and show all work.
-
(2 pts) Compute the coefficient C3,3 of the 8X8 DCT transform
matrix. Solve completely and show all work.
-
(2 pts) Reproduce the computations needed to derive the value -102.43 in
Table 13.6. You need to fully solve the calculations.
-
(3 pts) Estimate how many multiplications and additions are performed for
a DCT transform of a 4" X 6" color photograph
-
Assume a resolution of 150 dpi (dots or pixels per inch)
-
As an aside, most of today's digital cameras do these calculations AND
JPEG's quantization calculations AND JPEG's entropy encoding in a matter
of seconds!
-
For this answer, you may use the calculation approach discussed in class,
but note that in practice, more efficient calculation methods are used.
-
(2 pts) Encode the sequence "12 14 12 -35 -35 200" of JPEG DC values using
difference encoding and Table 13.9.
-
Note that Table 13.9 does not explain how the category is encoded.
Hence, your answer can be in pairs of "category bit-sequence" for
each DC value.
-
Assume an initial virtual DC value of 0
-
(2 pts) Encode only the AC components of the following transformed and
quantized 8x8 matrix:
-
7 2 4 -11 0
0 0 0
-
0 0 0 0 0
0 0 0
-
-3 -42 0 0 0
0 0 0
-
0 0 0 0 0
0 0 0
-
0 0 0 0 0
0 0 0
-
0 0 0 0 0
0 0 0
-
0 0 0 0 0
0 0 0
-
0 0 0 0 0
0 0 0
-
(2 pts) Use Table 13.10 to decode the following AC components. Show
your answer as an 8x8 matrix of integers.
Notes
-
Students may work individually or in groups of 2. Students
in different groups may NOT compare answers prior to submitting their work.
See the
syllabus for the full statement on the lateness and academic honesty policies.