Lecture 10: Modifying ImagesAnnouncementsGoals for TodaySimple Cosines in Space2D Frequency Content of Images Display Artifacts of the DCT (and DFT)Aliasing for ImagesAliasing for Images (cont)Another Aliasing ExampleSound vs. Image SP AnalogiesResizing ImagesAmplitude Scaling for ImagesReview: Grayscale vs. ColorColor Digital ImagesAdjusting Amplitude in Color ImagesChanging Color Images (cont.)Swapping Color PlanesReminder of Time Signal FilteringImage FilteringMore Filtering ExamplesAnother ExampleModifying Color ImagesNEW TOPIC: COMPRESSIONTypes of CompressionLecture 10: Modifying ImagesThe Digital World of MultimediaProf. Mari OstendorfEE299 Lecture 1030 Jan 2008Announcements HW3 due Friday in class Guest lecture Friday Feb 1 (EEB 403) A cultural history of JPEG… Dr. Joan Mitchell Another lecture by Dr. Mitchell on Thurs Patenting a wet suit… Thurs 1/31 10:30 EEB 125 Lab3: Read the lab *before* labEE299 Lecture 1030 Jan 2008Goals for Today Review of frequency content of images Sampling and aliasing for images Analogies in sound vs. image processing Image filteringEE299 Lecture 1030 Jan 2008Simple Cosines in SpaceEE299 Lecture 1030 Jan 20082D Frequency Content of Images DCT = Discrete Cosine TransformEE299 Lecture 1030 Jan 2008Display Artifacts of the DCT (and DFT)F=100300ppiF=160480ppiF=200600ppiEE299 Lecture 1030 Jan 2008Aliasing for ImagesF=50 for cosine, 300ppi originalEE299 Lecture 1030 Jan 2008Aliasing for Images (cont)Highest frequencies in center Æ aliasingRegion of aliasing grows with downsamplingThis is aliasing, not zooming in.EE299 Lecture 1030 Jan 2008Another Aliasing ExampleFor more interesting examples, see Orsak et al. pp. 133-134EE299 Lecture 1030 Jan 2008Sound vs. Image SP AnalogiesSound Time-scaling Amplitude scaling: change loudness Echo Time reverse Filter: LPF: smoothing, sounds more muffled HPF: detect onsets, sounds more tinnyImage Expand/shrink image Amplitude scaling: change brightness Double exposure Image flip Filter LPF: blurring, speckle removal HPF: deblurring, edge detectionEE299 Lecture 1030 Jan 2008Resizing Images Shrinking: Grayscale: replace KxK block with single pixel with average gray level of the K2 pixels in the original Color: Repeat for each color plane Enlarging: Grayscale: insert interpolated pixel values Color: repeat for each color plane, as in shrinkingEE299 Lecture 1030 Jan 2008+90-90Amplitude Scaling for ImagesEE299 Lecture 1030 Jan 2008Review: Grayscale vs. Color Gray: MxN matrix  Color: MxNx3B/W intensity:• 0 = black• 1 = whiteX(i,j)=0.4X(i,j,2)=0.6 X(i,j,3)=1.0R/G/B intensity:• [0,1]=[none, full]• (0,0,0)=black• (1,1,1)=whiteX(i,j,1)=0.758 bits/pixel Æ256 gray levels8 bits/color-pixel Æ 2563colors, 24 bits/pixelEE299 Lecture 1030 Jan 2008Color Digital ImagesComponent colorsEE299 Lecture 1030 Jan 2008Adjusting Amplitude in Color Imagespar=min(255,par+100);par(:,:,2)=min(255,par(:,:,2)+100);More of all colors Æ more whiteMore greenEE299 Lecture 1030 Jan 2008Changing Color Images (cont.)Turning off a color:X(:,:,1)=0;EE299 Lecture 1030 Jan 2008Swapping Color PlanesSwap:G Æ RB Æ GR Æ BoriginalEE299 Lecture 1030 Jan 2008Reminder of Time Signal Filteringoriginallow pass filtered(15 sample average)high pass filtered(avg of 5 samples – avg of next 5 samples)EE299 Lecture 1030 Jan 2008Image Filteringoriginal2D low pass filtered(average over 15x15 block)2D high pass filtered1D high pass filteredUsing the same filters as in the previous slide, but in 2D, e.g.filt2d=filt*filt’EE299 Lecture 1030 Jan 2008More Filtering Examplesoriginal2D low pass filtered 2D high pass filteredEE299 Lecture 1030 Jan 2008Another ExamplefilteredoriginalWhat type of filter was used?EE299 Lecture 1030 Jan 2008Modifying Color ImagesoriginalBlur only the green planeBlur everythingEach color plane can be manipulated as a grayscale image, e.g. filtered, contrast enhanced, etc.EE299 Lecture 1030 Jan 2008NEW TOPIC: COMPRESSION What is compression? Representing an signal (sound file, image or video) with fewer total bits Why bother? Storage: You can fit more songs & videos on your IPOD Communication: It takes less time to download stuff; real-time video conferencingEE299 Lecture 1030 Jan 2008Types of Compression Lossless Represent the sound/image/video with fewer bits without changing the signal *at all* (you can perfectly reconstruct the original) How? (like the 20 questions game) Map the different levels to a variable-length code, using short codes for more frequent things Lossy Represent the sound/image/video with fewer bits with some changes from the original How?  Throw out the stuff that people don’t pay attention