Vous êtes sur la page 1sur 5

Bi tp

1. Given the following image:

a) In the Hough transform, a point (x0, y0) in the (x y)-plane is mapped into the curve in the (,)-parameter space. Write down the equation of the curve. b) If we apply the Hough transform on the image below, what would be the maximum value for the accumulator cell in the (,) space. What the corresponding (,) value? Note: each black square denotes a point and the numbers are in coordinates. 2. The two texture images shown below are quite different, but their histograms are identical. Both images have size 80x80, with white (1) and black (0) pixels.

a) Suppose that both images are blurred with 3x3 smoothing mask. Would the resultant histograms still the same? Draw the two histograms and explain your answer. Note: The black lines are used to signify the boundaries of the two images but not part of them. b) For images (b), computer the co-occurrence matrix corresponding to the property a pixel to the right. Each gird (black or white) in image (b) is of size 10x10. 3. Given the following image:

a) Give a sequence of f (xi, yi) lying on the object boundary, briefly describe how to obtain its Fourier descriptor? Why the Fourier descriptor capable of dealing with noisy boundaries?

b) As discussed in the lecture notes, one can obtain two types of signatures based on the object boundary. One of them is based on distances and the other is based on angles. Describe how to obtain them. c) A square boundary in the domain [0,1]x[0,1] is shown below. Draw the two signatures obtained in Question 3b. 4. Given the following image:

a) Discuss the advantages and disadvantages of global and local histogram equalization. b) The following figure shows - a 3-bit image of size 5-by-5 image in the square, with x and y coordinates specified - a Laplacian filter and. - a low pass filter. Compute the following: a) b) c) d) e) f) The output of a 3x3 mean filter at (2,2) The output of a 3x3 median filter at (2,2) The output of a 3x3 Laplacian filter shown above at (2,2) The output of a 3x3 low pass filter shown above at (2,2) The histogram of the whole image. The result of histogram equalization at the point (2, 2). Show steps in obtain your solution.

5. Answer the questions: a) Explain the image degradation model described in the lecture notes, and how to use inverse filter for image restoration. b) Inverse filter many encounter numerical problem in practice. Provide one solution to overcome this instability. 6. For the following boundary (in counterclockwise direction)

Computer the chain code that is:

a) Invariant to starting point; b) Invariant to both starting point and rotation. 7. Give a gray image I, we want to find all its pixels that are in the domain [a, b]. Design a matlab function F=f(I, a, b) that returns a binary matrix F that is of the same size as I, where 1 for pixels satisfying the domain condition and 0 otherwise. 8. State whether the following statement is True (T) or False (F), incorrect answers may cancel (delete) correct a. Lumen is a unit of light flow. b. Sensor strips mounted in a ring configuration are used in medical and industrial imaging to obtain cross- sectional (slice) images of 3-D objects. c. An image with low contrast has wide histogram distribution. d. The idea behind contrast stretching is to increase the dynamic range of the gray levels in the image being processed. e. High-boost filtering gives us the flexibility to decrease the contribution made by the image to over all enhanced result. f. A Degradation is defined at point for which H(u, v) is down to a certain fraction of its maximum value. g. Convolution is same as correlation expect that, the filter is first rotated by degree. h. Power-law function maps a narrow range of low gray-level values in the input image into a power a wider range of output levels. i. Single valued is required to guarantee the inverse transformation will exist. j. Laplacian filter replaces the pixel value by the median value in the neighborhood. Short answer questions What is the definition of image? Define the Bit-Plane slicing If the pixels in an image are shuffled (reordered), will there be any change in the histogram? Justify your answer. d. What is an ideal low-pass filter? Is this filter suitable to use in terms of image processing? 10. Suppose that you have been given the 3-bit 4x4 image shown in the figure below. 9. a. b. c.

a. Show the output image as a result of histogram equalization.

b. Using the same image matrix as in the previous example, show the output image as the desired histogram is as follows (10 pts)

11. Given an image of size 3x3 as f(x,y), determine the output image g(u,v) using Logarithmic transformation , ) | 1 , )) by choosing c as 1) c = 1 2) c = L/ (log10 (1+L))

12. Filter the given 4x4 gray level image with a. 3x3 mean filter using zero padding. b. Laplacian filter with given mask and reflecting the border pixels.

0 1 1 4 0 1

0 1 0

13. Given an input image of size 7x7 shown below, was filtered using 3x3 adaptive median filter with maximum allowed size of 5x5

a. What is the value of the pixel A in the input image?

b. What are the values of the pixels x, y and z in the input image? Pixel value A x y z 14. In the figure below, when the images on the left was filtered was using a smoothing filter, the result was the image on the right. The filter used was one of these: Averaging filter. Ideal low pass filter Gaussian low pass filter Median filter The small black square on the lower right hand corner of the original image shows the size of the mask that was used. That small square is not part of the image.

a) For each of the four possible filters listed above, give at least one reason why you think it was, or was not, the filter actually used. b) If the size of the mask were tripled, and the same filter you selected in (a) used, how would the appearance of the image after filtering be changed?

Vous aimerez peut-être aussi