Sampling and Reconstruction Digital Image Processing How does it - - PowerPoint PPT Presentation

Sampling and Reconstruction

Digital Image Processing

Slide 2 ICS 280: Visual Perception

How does it help?

Input (256 x 256) Subsampled(128 x 128) Subsampled from filtered image(128 x 128) Insufficient sampling. Hence, aliasing. Filtering reduces frequency content. Hence, lower sampling is sufficient. Filtered (256 x 256) ANTI-ALIASING

Slide 3 ICS 280: Visual Perception

Sampling – Frequency Domain

f f s

s

2

• f
• f s

s

2 f f s

s

2 f 2 f s

s

3 f 3 f s

s

• f
• f s

s

f f s

s

2 f 2 f s

s

3 f 3 f s

s

• f
• f s

s

Convolution

Slide 4 ICS 280: Visual Perception

Reconstruction – Frequency Domain

f f s

s

2

• f
• f s

s

2 f f s

s

2 f 2 f s

s

3 f 3 f s

s

• f
• f s

s

f f s

s

2

• f
• f s

s

2

Multiplication

Slide 5 ICS 280: Visual Perception

Reconstruction (Wider Kernel)

f f s

s

2

• f
• f s

s

2 f f s

s

2 f 2 f s

s

3 f 3 f s

s

• f
• f s

s

f f s

s

2

• f
• f s

s

2

PIXELIZATION: Lower frequency aliased as high frequency

Slide 6 ICS 280: Visual Perception

Reconstruction (Narrower Kernel)

f f s

s

2

• f
• f s

s

2 f f s

s

2 f 2 f s

s

3 f 3 f s

s

• f
• f s

s

f f s

s

2

• f
• f s

s

2

BLURRING: Removal of high frequencies

Slide 7 ICS 280: Visual Perception

Aliasing artifacts (Right Width)

Slide 8 ICS 280: Visual Perception

Wider Spots (Lost high frequencies)

Slide 9 ICS 280: Visual Perception

Narrow Width (Jaggies, insufficient sampling)