Contents
Preface
xiii
Acknowledgements
xvii
The Representation of Digital Images
DSP Chips and Image Processing
Useful Internet Resources
1
2
3
4
6
9
10
13
The TMS320C6000 Line of DSPs
15
16
Introduction
1.1
1.2.
1.3
1.4
1.5
1.6
1.7
Structure and Organization of the Book
Prerequisites
Conventions and Nomenclature
CD-ROM
2. Tools
2.1.
2.1.1
2.1.2
2.1.3
VLIW and VelociTI
Fixed-Point versus Floating-Point
TI DSP Development Tools
(C6701 EVM & C6416 DSK)
2.2
TI Software Development Tools
2.2.1 EVM support libraries
2.2.2 Chip Support Library
2.2.3 DSP/BIOS
2.2.4 FastRTS
2.2.5 DSPLIB and IMGLIB
17
21
25
26
28
28
29
29
29
x
Contents
2.3
MATLAB
2.4.
Visual Studio .NET 2003
2.4.1 Microsoft Foundation Classes (MFC)
2.4.2 GDI+
2.4.3 Intel Integrated Performance Primitives (IPP)
3. Spatial Processing Techniques
Spatial Transform Functions and the Image
Histogram
3.2
Contrast Stretching
3.2.1 MATLAB Implementation
3.2.2 TI C67xx Implementation and MATLAB Support Files
3.3
Window/Level
3.3.1 MATLAB Implementation
3.3.2 A Window/Level Demo Application Built
Using Visual Studio .NET 2003
3.3.3 Window/Level on the TI C6x EVM
3.4
Histogram Equalization
3.4.1 Histogram Specification
3.4.2 MATLAB Implementation
3.4.3 Histogram Specification on the TI C6x EVM
30
30
31
33
34
37
3.1
4. Image Filtering
4.1
Image Enhancement via Spatial Filtering
4.1.1
4.1.2
4.1.3
4.1.4
Image Noise
2D Convolution, Low-Pass and High-Pass Filters
Fast Convolution in the Frequency Domain
Implementation Issues
4.2
Linear Filtering of Images in MATLAB
4.3
Linear Filtering of Images on the TI C62xx/C67xx
4.3.1 2D Filtering Using the IMGLIB Library
4.3.2 Low-Pass Filtering Using DSPLIB
4.3.3 Low-Pass Filtering with DSPLIB and Paging
4.3.4 Low-Pass Filtering with DSPLIB and Paging via DMA
4.3.5 Full 2D Filtering with DSPLIB and DMA
4.4
Linear Filtering of Images on the TI C64x
4.4.1 Low-Pass Filtering with a 3x3 Kernel Using IMGLIB
4.4.2 A Memory-Optimized 2D Low-Pass Filter
4.5
Non-linear Filtering of Images
4.5.1 Image Fidelity Criteria and Various Metrics
4.5.2 The Median Filter
4.5.3 Non-Linear Filtering of Images in MATLAB
4.5.4 Visual Studio .NET 2003 Median Filtering Application
37
40
45
47
59
63
68
77
82
87
90
94
103
103
106
109
111
114
116
118
120
124
128
132
138
139
141
146
152
152
155
159
166
Embedded Image Processing on the TMS320C6000 DSP
4.5.4.1
Generating Noise with the Standard C Library
4.5.4.2
Profiling Code in Visual Studio .NET 2003
4.5.4.3
Various Median Filter C/C++ Implementations
4.5.5 Median Filtering on the TI C6416 DSK
4.6
Adaptive Filtering
4.6.1 The Minimal Mean Square Error Filter
4.6.2 Other Adaptive Filters
4.6.3 Adaptive Image Filtering in MATLAB
4.6.4 An MMSE Adaptive Filter Using the Intel IPP Library
4.6.5 MMSE Filtering on the C6416
5. Edge Detection and Segmentation
5.1
Edge Detection
5.1.1
5.1.2
Edge Detection in MATLAB
An Interactive Edge Detection Application with
MATLAB, Link for Code Composer Studio, and RTDX
5.1.2.1
DSP/BIOS
5.1.2.2
C6416 DSK Target
5.1.2.3
C6701 EVM Target
5.1.2.4
Host MATLAB Application
5.1.2.5
Ideas for Further Improvement
5.2
Segmentation
5.2.1 Thresholding
5.2.2 Autonomous Threshold Detection Algorithms
5.2.3 Adaptive Thresholding
5.2.4 MATLAB Implementation
5.2.5 RTDX Interactive Segmentation Application with
Visual Studio and the TI C6416
5.2.5.1
C6416 DSK Implementation
5.2.5.2
Visual Studio .NET 2003 Host Application
6. Wavelets
6.1
Mathematical Preliminaries
Quadrature Mirror Filters and Implementing the
2D DWT in MATLAB
6.1.2 The Wavelet Toolbox
6.1.3 Other Wavelet Software Libraries
6.1.4 Implementing the 2D DWT on the C6416 DSK
with IMGLIB
6.1.4.1
Single-Level 2D DWT
6.1.4.2
Multi-Level 2D DWT
6.1.4.3
Multi-Level 2D DWT with DMA
6.2
Wavelet-Based Edge Detection
xi
167
169
171
179
183
185
187
190
194
198
211
212
219
225
229
232
236
240
245
246
248
249
254
257
261
262
269
281
282
6.1.1
287
296
299
299
301
305
309
313
xii
Contents
62A
6.2.2
The Undecimated Wavelet Transform
Edge Detection with the Undecimated
Wavelet Transform
6.2.3 Multiscale Edge Detection on the
C6701 EVM and C6416 DSK
6.2.3.1 Standalone Multiscale Edge Detector (C6701 EVM)
6.2.3.2 HPI Interactive Multiscale Edge Detector Application
with Visual Studio and the TIC6701 EVM
6.2.3.2.1 C6701 EVM Target
6.2.3.2.2 Visual Studio .NET 2003 Host Application
6.2.3.3 Standalone Multiscale Edge Detector (C6416 DSK)
6.3
Wavelet Denoising
6.3.1 Wavelet Denoising in MATLAB
6.3.2 Wavelet Denoising on the C6x
6.3.2.1
D4 DWT and EDWT functions on the C6416
6.3.2.2
A C6416 Wavelet Denoising Implementation
Putting it Together: A Streaming Video
Application
A. 1
Creation and Debugging of MEX-files
in Visual Studio .NET 2003
A. 1.1 The i m p o r t _ g r a y s c a l e _ i m a g e MEX-file
A. 1.2 A MEX-file for HPI communication between
MATLAB and the C6x EVM
A.2
The C6701 EVM Program
A.3
MATLAB GUI
A.4.
Ideas for Further Improvement
316
317
323
323
329
331
334
341
347
352
359
362
374
Appendix A
379
382
385
390
393
396
398
Appendix B Code Optimization
B.I
Intrinsics and Packed Data Processing
B. 1.1 Packed Data Processing
B. 1.2 Optimization of the Center of Mass Calculation
on the C64x Using Intrinsics
B.2
Intrinsics and the Undecimated Wavelet Transform
B.3
Convolution and the DWT
401
404
405
408
415
418
Index
425
">
