Image Processing Reference
In-Depth Information
(a)
(b)
(c)
(d)
(e)
(f)
(g)
Fig. 2.7
Resized images: (I) Garden: (a) Original image, (b) Nearest neighbor, (c) Bilinear,
(d) Bicubic, (e) NEDI, (f) Edge-oriented and (g) CA based method
Ta b l e 2 . 2
Execution time (msec). NN: Nearest; BL: Bilinear; BC: Bicubic; ND: Nedi, EO:
Edge - oriented; CA-R method
Image
Method
NN
BL
BC
ND
EO
CA-R
Koala (RGB: 256
×
192)
3.6
19
21
44860
91.2
84.4
Cameraman (GR:128
×
128)
3.1
8.7
9.7
4500
38.2
36.9
Lena (RGB:150
×
150)
3.9
13.2
15.9
18600
41.2
39.3
Box (GR:320
×
240)
3.6
9.4
10.6
5000
63.2
58.6
Building (RGB:640
×
480)
5
26.2
29.5
65290
102.4
99.4
Teddy (RGB:450
×
376)
5.052
9.829
16.002
10526
61.235
59.331
Statue (RGB:384
×
288)
4.878 12.574 10.083
6537
43.216
43.808
Butterfly (RGB:133
×
100)
4.395
6.493
9.228
2930
12.648
13.633
Port (RGB:800
×
600)
6.006 22.002 34.731 128494 152.367 169.085
×
Garden (RGB:352
240)
4.157 11.445 12.894
21507
34.504
36.135
2.5
Hardware Implementation
The proposed model was implemented on an FPGA platform. Although the model
performed well as software on conventional, commercially available computers,
there are many reasons why a specialized, custom made integrated circuit could be
preferred in the case of CA parallel computation tool. Conventional general-purpose
computers display two main disadvantages: large sizes and increased operational
energy amounts. On the other hand smaller, embedded general-purpose processors
may pose limitations on the computational complexity of the simulated model. The
Search WWH ::
Custom Search