EE 535 Digital Image Processing Sample Questions for Final Exam 1. Given the degradation function as follows: g = f*h + n where g is degraded image, h is convolution filter, f is latent image and n is noisy. Describe a method to estimate f if we don’t know about h. 2. Describe the following equation: 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. Discuss how Poisson image editing can achieve seamless cloning. Suggest a method to calibrate camera response function. Suggest a method to resize image without changing aspect ratio of important objects Describe the algorithm to apply the Hough transform for line detection What are the differences between lossless compression and lossy compression? Describe the process to get an HDR image from LDR images. What is Bilateral filter? Discuss the effects of parameters in Bilateral filter. Describe the differences between Gaussian filter and Bilateral filter Compare the difference between the first and the second image derivative What is Huffman coding? Use Huffman encoding to get the optimal variable-length code of the following symbols. What is the average number of bits code length of the optimal variable-length code. Original source Symbol a1 a2 a3 a4 a5 a6 14. 15. 16. 17. 18. 19. Probability 0.03 0.5 0.18 0.09 0.1 0.1 Source Reduction 1 2 3 4 What is Gamma correction? Describe histogram equalization. Describe inverse filter and Wiener filter. What is ringing artifact? How to reduce the ringing artifact. Describe Radon Transform Describe a feature that can be used in video compression, but cannot be used in image compression. 20. Please give an example where two images have exactly the same histograms in RGB space, but have different histogram in HSI space. Assume the histograms are plotted individually for each color channel. 21. Refers to the graph below, describe the effects to region A, and region B. 22. Apply Sobel, Laplacian, and median filter to the below 5x5 matrix. Use zero padding for boundary. 1 10 11 20 21 2 9 12 19 22 3 8 13 18 23 4 7 14 17 24 5 6 15 16 25 23. What can you tell from the slope of intensity manipulation curve? Please describe the effect between [R1,R2] and [R2,R3]. 24. Why do we want to have a monotonic curve for intensity manipulation? 25. What is the effects of this function I(x) = I(x)*I(x) for image intensity manipulation, assume image intensity is between zero and one? 26. What are the differences between Global Histogram Equalization and Local Histogram Equalization? Please write down two differences. 27. What is the effect of using Gaussian Filter with larger neighborhood size? 28. Describe the differences between mean filter and median filter. 29. Discuss why the summation of weights in smoothing filter (i.e. mean filter, Gaussian filter) needed to be equal to 1. 30. is the equation for finding x- and y-derivative, write down the equation to compute the magnitude and direction of gradients. 31. What would happen to the second derivative filter if there are image noises? And what is a solution for this problem? 32. Histogram Equalization is used to enhance image contrast and to balance image brightness. Suppose you have the following histogram, what would be the output after histogram equalization? Please draw the cumulative distribution function and output histogram. 33. Beside histogram equalization, another way to enhance image contrast is to find the maximum and minimum intensity of an image and then stretch the intensity range by mapping the maximum and minimum intensity of an image to 0 and 255. Please write down the equation for this manipulation. 34. Describe the effect of the following equation: 35. Why there are image noises? Please give two reasons and describe briefly how it appears in image. 36. Describe the steps on how to estimate the noise distribution parameters from image. 37. Suppose you know the model of image noises, can you recover the original noise-free image? Why? 38. Please write down the image degradation model for motion blur. 39. Why image deconvolution is difficult? Please give two reasons. 40. In wiener filter, how to estimate the parameter for SNR? What are the effects of this parameter? 41. Why corner is a good feature for image analysis? 42. Describe two features that can be used for salient region detection. 43. Why do we need to segment an image? Please give 3 reasons. 44. The simplest image segmentation algorithm is based on grouping of color intensity. What information, other than colors, can be used for image segmentation? Please give 2 examples. 45. Describe the procedures of K-mean clustering algorithm for image segmentation. 46. What are the differences between EM algorithm and K-mean clustering algorithm for image segmentation? 47. Describe the procedure to use Hough transform for circle detection in images. Assume that you know the radius of circles. 48. Describe two examples of human gestalt rules that can be used for segmentation and grouping. 49. What is the effect of sigma in Canny edge detector 50. Given the following image and 3x3 Sobel operator, convert the image into a 1D vector form, and write down the operation matrix which corresponds to the given 3x3 Sobel operator. Use “circular” to handle boundary. 3 4 5 7 2 4 6 5 5 7 7 4 5 5 6 Image 51. Describe the following graph: -1 0 1 -2 0 2 1 0 1 Sobel operator 52. What is order statistic filter? Please give one example and describe its usage. 53. Describe the cause of image degradation, and discuss how to restore an image suffered from the degradation.
© Copyright 2019