Image Processing Basics - Digital Signal Processing: Fundamentals and Applications - page 758

improve the image contrast and effectively enhances contrast for image underexposure. Color
image equalization can be done only in the luminance channel or RGB channels.
6. Image enhancement techniques such as average lowpass filtering can filter out random noise in
the image; however, it also blurs the image. The degree of blurring depends on the kernel size.
The bigger the kernel size, the more blurring occurs.
7. Median filtering effectively removes the “pepper and salt” noise in an image.
8. The edge detection filter with Sobel convolution, Laplacian, and Laplacian of Gaussian kernels
can detect image boundaries.
9. The grayscale image can be made into a facsimile of the color image by pseudo-color image
generation, using the red, green, and blue transformation functions.
10. RGB-to-YIQ transformation is used to obtain the color image in YIQ space, or vice versa . It can
also be used for color-to-grayscale conversion, that is, keeping only the luminance channel after
the transformation.
11. 2D spectra can be calculated and are used to examine filtering effects.
12. JPEG compression uses the 2D-DCT transform for both grayscale and color images in the YIQ
color space. JPEG uses different quality factors to normalize DCT coefficients for the luminance
(Y) channel and the chrominance (IQ) channels.
13. The mixing of two images, in which two pixels are linearly interpolated using the weights 1 a
and a , can produce video sequences that have effects such as fading in and fading out of images,
blending of two images, and overlaying of text on an image.
14. Analog video uses interlaced scanning. Avideo frame contains odd and even fields. Analog video
standards include NTSC, PAL, SECAM.
15. Digital video carries the modulated information for each pixel in the YCbCr color space, where Y
is the luminance and Cb and Cr are the chrominance. Chroma subsampling creates various digital
video formats. The industry standards include CCIR601, CCR601, CIF, and QCIF.
16. Motion compensation of a video sequence produces motion vectors for all the image blocks in the
target video frame, which contain displacements of these image blocks relative to the reference
video frame. Recovering the target frame involves simply copying each image block of the
reference frame to the target frame at the location specified in the motion vector. Motion
compensation is a key element in MPEG video.
14.12 PROBLEMS
14.1. Determine the memory storage requirement for each of the following images:
a. 320 240 8-bit grayscale
b. 640 480 24-bit color image
c. 1600 1200 8-bit indexed image
14.2. Determine the number of colors for each of the following images:
a. 320 240 16-bit indexed image
b. 200 100 24-bit color image