Unit 5 - Practice Quiz

1 Why is color information highly valuable in computer vision applications?

2 Which of the following tasks benefits the most from adding color information to a standard grayscale image?

3 Which color model is primarily based on how the human eye perceives colors and is standard for displaying images on digital screens?

4 What do the letters in the HSV color model stand for?

5 Why is it common to convert images from RGB to the HSV color space in computer vision?

6 When converting an RGB image to a grayscale image, what mathematical operation is typically performed on the pixels?

7 What is the primary purpose of applying color augmentation techniques to an image dataset?

8 Which of the following is a common color augmentation technique?

9 What does 'color constancy' refer to in computer vision?

10 Which simple algorithm is often used as a baseline to achieve color constancy by assuming that the average color of a scene is neutral gray?

11 What does the pixel value represent in a range image?

12 What is another common name for a range image?

13 Which of the following best describes the difference between a 2D intensity image and a 3D range image?

14 Which type of image is generally immune to shadows and variations in ambient lighting?

15 How do active range sensors operate?

16 Which of the following is an example of an active range sensor?

17 Why is preprocessing often necessary for raw range data?

18 Which filter is commonly used in preprocessing range images to remove outlier noise (salt-and-pepper noise) without blurring the sharp geometric edges?

19 Which of the following fields heavily relies on range data for obstacle avoidance and mapping?

20 How are range images primarily used in modern smartphone facial recognition systems (like FaceID)?

21 Why is color information often considered more robust than grayscale for object tracking under partial occlusion?

22 In image segmentation, how does the inclusion of color channels typically affect the performance of clustering algorithms like K-means compared to using only intensity?

23 Which color model is most suitable for developing an application that requires intuitive manipulation of "tint" and "vividness" by a human user?

24 In the YCbCr color model used in video compression, why is the 'Y' channel typically sampled at a higher resolution than 'Cb' and 'Cr'?

25 When converting from an RGB image to grayscale using the standard NTSC formula , why are the weights unequal?

26 What happens if the non-linear gamma correction step is omitted during the conversion from linear RGB to sRGB space?

27 When applying color jittering as an augmentation technique to train a robust neural network, which parameter should be perturbed to simulate varying lighting intensities without changing the object's intrinsic color?

28 What is the primary purpose of applying PCA-based color augmentation (like Fancy PCA) on RGB images during deep learning model training?

29 Which algorithm applies the assumption that the spatial average of surface reflectances in a scene is achromatic (gray) to achieve color constancy?

30 An image taken under a strong yellow tungsten light looks unnaturally warm. How does the White Patch (Max-RGB) algorithm attempt to correct this illumination bias?

31 In a standard range image, what does the scalar value at pixel coordinate explicitly represent?

32 Which of the following best describes the structural representation of a standard range image?

33 Why are edge detection algorithms applied to range images fundamentally different in physical interpretation compared to 2D intensity images?

34 How does the presence of ambient illumination changes affect 2D intensity images compared to 3D range images?

35 How does a Time-of-Flight (ToF) camera estimate the depth of a scene?

36 In a structured light active range sensor, what is the primary purpose of projecting a known pattern (e.g., a grid or stripes) onto the scene?

37 Range images obtained from active sensors often contain "flying pixels" (mixed pixels) at object boundaries. Which filter is most effective for removing these artifacts while preserving sharp depth edges?

38 What is a common technique used to fill in missing depth values (holes) in range images caused by highly reflective surfaces or occlusions?

39 In autonomous navigation, how is range data primarily utilized for obstacle avoidance?

40 How are range images utilized in industrial quality inspection for manufactured parts?

41 Metamerism poses a significant challenge in color-based object recognition. Two physically distinct surfaces are considered metamers under a specific illuminant if they exhibit which of the following properties?

42 According to the Dichromatic Reflection Model, the total radiance from a dielectric inhomogeneous material is given by . If an algorithm successfully factors out , what is the primary consequence for computer vision tasks?

43 The sRGB color space applies a non-linear transfer function (gamma correction) to linear RGB values. From a mathematical and information-theoretic perspective, what is the primary purpose of applying this specific non-linearity prior to 8-bit quantization?

44 In the YCbCr color model, chroma subsampling (e.g., 4:2:0) reduces bandwidth. Which mathematical property of the RGB to YCbCr conversion makes this subsampling perceptually viable?

45 When converting from linear RGB to the HSI (Hue, Saturation, Intensity) color space, the Hue becomes mathematically undefined when . How does this singularity impact gradient-based optimization in deep learning models operating directly in HSI space?

46 The conversion from CIE XYZ to CIE Lab* incorporates a non-linear function (for ). What is the exact geometrical objective of this specific transformation?

47 In Fancy PCA (used in AlexNet) for color augmentation, an RGB pixel is augmented by adding , where and are eigenvectors and eigenvalues of the RGB covariance matrix. What specific invariance is this technique injecting into the model?

48 When applying color jittering (randomly varying Hue, Saturation, and Value) in HSV space to augment data, which mathematical precaution MUST be taken for the Hue channel that is not required for Saturation or Value?

49 The Grey-World assumption is a popular algorithm for color constancy. Under which of the following real-world scenarios will the Grey-World algorithm fail most catastrophically, producing severely distorted colors?

50 The von Kries transform models illuminant changes using a diagonal matrix such that . For this diagonal approximation to be strictly true mathematically (without error), what underlying constraint must be satisfied regarding the camera sensors?

51 In a range image, pixel values represent distance from a reference plane or sensor. If a range image is generated using perspective projection, how does the spatial resolution (the physical area covered by a single pixel) change as the depth value increases?

52 When computing geometric properties, how does the evaluation of Gaussian Curvature on a 3D range image fundamentally differ from applying edge detection (e.g., Canny) on a 2D intensity image?

53 A key challenge in 2D computer vision is 'scale ambiguity', which is generally resolved in 3D range images. However, 3D range imaging suffers from a distinct physical limitation not typically affecting ideal 2D pinhole cameras. Which of the following is it?

54 In an Amplitude Modulated Continuous Wave (AMCW) Time-of-Flight sensor, the distance is calculated using the phase shift of the modulated signal at frequency . The measurement is subject to phase wrapping. What is the expression for the unambiguous range interval ?

55 Structured light scanners project known patterns to establish corresponding points. In a highly cluttered scene with strong inter-reflections, a standard binary Gray code projection often fails. Why is phase-shifting (using sinusoidal patterns) often combined with Gray codes to mitigate this?

56 When smoothing a noisy range image (depth map), applying a standard linear Gaussian filter causes severe blurring across object boundaries (depth discontinuities). The Bilateral Filter solves this by modifying the convolution kernel. How does the Bilateral Filter achieve this mathematically?

57 In the Iterative Closest Point (ICP) algorithm used for range data registration, using a 'point-to-plane' error metric instead of a 'point-to-point' metric generally results in which of the following optimization behaviors?

58 'Flying pixels' are artifacts in ToF range images occurring at depth discontinuities, caused by the sensor integrating light from both foreground and background objects within a single pixel. Which morphological filter approach is most suited to systematically identify and remove these without eroding legitimate thin structures?

59 To estimate the surface normal at a specific point in a 3D point cloud (derived from range data), one standard method involves applying Principal Component Analysis (PCA) to the local -nearest neighbors. What does the eigenvector corresponding to the smallest eigenvalue of the local covariance matrix represent?

60 In 3D object recognition from range data, a 'Spin Image' is a local descriptor. How does the Spin Image achieve invariance to rigid 3D transformations (rotations and translations)?