Image Processing Questions

• Image Formation

  • What is color constancy?

    What is the purpose of color constancy?

    What determines the color a human observer perceives in a scene?

    What is the definition of color?

    What is the Maxwell-Hemlholtz experiment?

    Is there a 1-1 correspondence between pure colors and monocromatic light?

    What are rods and cones?

    Cones are labeled R, G, and B; what do those labels mean and are they accurate?

    What is the "line of purples"?

    What is metamerism?

    What is an "unrealizable color"?

    What is color gamut?

    What is a perceptual color space?  How is it constructed?

    What is the HLS color space?

    What is the camera model for a pinhole camera?

    What is a Bayer pattern?

    What is intensity quantization in digital sensors?  What is the consequence?

    What is spatial quantization?  What is the consequence?

    What is an iso-preference curve?

    Name four major image formats, describe what image types they are for, and what compression methods they use.

• Pixel Operations and Linear Filters
  • What is a linear filter?
  • Give an example of a non-linear filter.
  • Prove that the ______ filter is linear/non-linear.
  • What is a FIR filter?
  • What is an IIR filter?
  • What is a Gaussian filter?
  • What is a Laplacian of Gaussian filter?
  • What is a Derivative of Gaussian filter?
  • What is an unsharp mask filter?
  • Express convolution as matrix multiplication.
• Fourier Transform
  • State the 1D discrete Fourier transform.
  • State the 1D continuous Fourier transform.
  • State the Fast Fourier Transform algorithm.
  • What is the complexity of the FFT?
  • What is the Thomasson-Lanczos-Cooley-Tukey-Gauss Lemma?
  • What is the Fourier transform of the convolution of two signals?
  • How can you use the convolution theorem to speed up convolutions?  What is the resulting complexity?
  • What is the Fourier transform of a Gaussian?  Derive it and its mean and variance.
  • Derive the Fourier transform of a signal that has been translated in time by an amount d.
  • What is a short time Fourier transform?
  • What is a window function?
  • Why do you need a window function?
• 2D Fourier Transform

  • State the 2D discrete/continuous Fourier transform.

    Is the Fourier transform its own inverse?

    What is separability?

    How does separability affect computational complexity?

    Given an image corrupted by a 2D sinusoid, how can you use the 2D Fourier transform to remove the sinusoid?

    How much (complexity) can you speed up 2D convolutions by using the FFT and separability?

    What is the DCT?

    How do you use the DCT for image compression?

    Why does image compression use a zig-zag traversal of DCT coefficients?

    How is parameter quantization used in DCT compression?

    What is a run-length code?  How is it used in DCT-based image compression?

    What is aliasing?  What is an anti-aliasing filter?

    Give an example of a wavelet.

    Describe how wavelets can be used for image compression.

• Mathematical Morphology

  • Name the basic operations of mathematical morphology.

    Give mathematical definitions of the basic operations of mathematical morphology.

    What is a "bit blit" and how can you implement mathematical mophology in terms of it?

    What is the brushfire algorithm?

    What is the hit-or-miss transform?

    What is morphological edge detection?

    How would you remove noise from this _____ noisy image using mathematical morphology?

    Explain thinning algorithms.

    Describe the 2-pass distance transform algorithm.

    Describe the queue based Euclidean distance transform algorithm.

    Prove this ___________ identity in mathematical morphology.

    Smoothing with a Gaussian followed by thresholding performs an operation that is similar in some ways to opening/closing.  What are the similariies and differences?

• Matching
  • What is template matching?
  • Describe and explain the Hough transform algorithm.
  • Describe and explain the RANSAC algorithm.
  • Describe and explain the RAST algorithm.
• Segmentation
  • What is the relationship between segmentation and thresholding algorithms?
  • Describe color image segmentation using the k-means algorithm.
  • Describe and explain the watershed image segmentation algorithm.
  • Explain edge-based image segmentation.
  • Explain segmentation by region growing.
  • Describe and explain MRF-based image segmentation.
• Binary Image Analysis

  • Name common descriptors for objects in binary images.

    Explain how boundary tracing works.

    Explain how boundaries can be represented as sequences of complex numbers, and why you would want to do that.

    Explain the relationship between moments, major and minor axes of ellipses, and orientations of binary objects.

    What is a chain code?

• Edge Detection
  • What is the motivation for edge detection from a psychological point of view?
  • What are some engineering motivations for edge detection?
  • Describe the steps of the Canny edge detector.
  • Describe the design criteria of the Canny edge detector.
  • Describe how to compute polygonal approximations.
  • Describe the criteria for ridge detection.