Spatial Filtering

Question 1

What is image processing?

Answer 1

Image processing is a technique of applying a relevant mathematical operation or an algorithm on a digitized image to generate an enhanced image or extract features.

Question 2

What are the different types of image processing?

Answer 2

• Image enhancement (Eg brightness reduction)
• Image Restoration
• Colour image processing (Green screens)
• Morphological operations
• Image segmentation
• Object detection
• Image blurring
• Edge detection

Question 3

What is the purpose of spatial filtering?

Soothing, removal,

Answer 3

• Image smoothing
• Noise removal
• Image sharpening
• Edge detection
• Inpainting
• Pre-processing

Question 4

What are the steps for spatial filtering?

Answer 4

• Define filter
• Identify the image neighbourhood
• Multiply each pixel value in the neighbourhood with the corresponding weight in the filter mask and sum the result.
• Shift one pixel to the right and repeat step 3.
• Repeat until done

Question 5

Why in spatial filtering does the weight have to sum to 1?

Answer 5

To preserve the overall brightness

Question 6

Why in spatial filtering does the weight have to sum to 1?

Answer 6

To preserve the overall brightness

Question 7

What happens in spatial filtering if the weight is above 1?

Answer 7

The image will be brighter

Question 8

For the averaging filter, what does a larger mask mean?

Answer 8

More blurring

Question 9

What is the purpose of the unsharp mask?

Answer 9

• The idea is to improve the sharpness of the edges.
• The name is from the fact that the process involves blurring the original image.

Question 10

What filter does the unsharp mask use?

Answer 10

Averaging filter

Question 11

What is J?

Answer 11

Output image

Question 12

What is I?

Answer 12

Original image

Question 13

How do you determine J in terms of the unsharp mask?

Answer 13

J = I + A x [I - Iblur]
A = scaling factor

Question 14

What are examples of linear spatial filters?

Answer 14

• Larger average filter (smoothing filter)
• Vertical difference (horizontal edge detection)
• Horizontal difference (vertical edge detection)
• Horizontal and vertical edge detection
• Diagonal difference filter

Question 15

What is a median filter?

Answer 15

• Similar to linear filtering but instead of calculating a weighted sum of the pixels within each neighbourhood, the output pixel value is determined as the neighbourhood median.
• Very effective at removing a certain type of image noise

Question 16

How does the non-local means filter work?

Answer 16

• Works by finding a small region of an image and other similar patches on the image which lies in the proximity of the patch in question and averages them together.
• Assuming the noise is random in its location.

Question 17

What does the non-local mean filter fix?

Answer 17

• Helps to reduce or suppress the noise
• Adding more patches together reduces the noise further.
• Removes speckles
• Not at the expense of reducing the sharpness of the image.
• Preserve the edges