Lecture 7: Sounds & Imagesssli.ee.washington.edu/courses/ee299/notes/lecture7.pdf · Sounds &...

14
Lecture 7: Sounds & Images The Digital World of Multimedia Prof. Mari Ostendorf

Transcript of Lecture 7: Sounds & Imagesssli.ee.washington.edu/courses/ee299/notes/lecture7.pdf · Sounds &...

  • Lecture 7: Sounds & Images

    The Digital World of MultimediaProf. Mari Ostendorf

  • EE299 Lecture 723 Jan 2008

    AnnouncementsAnother office hour change:

    Friday change to 10-12; Thurs 1:30-3:00Problem with Lab dropbox is fixed – try againThis Friday: Guest Lecture

    Color VisionProf. Maya Gupta, UW EE

    Tomorrow EE lecture of interestRetinal local adaptation for the tone mapping of color filter array images

    Prof. Sabine Susstrunk, EPFL1/24, 4pm, EEB 403

  • EE299 Lecture 723 Jan 2008

    Goals for TodayReview: FilteringCollaborative quizSignal generationIntro to Digital Images

  • EE299 Lecture 723 Jan 2008

    Filtering Review

    Low Pass Filtered High Pass Filtered

    Original Signal

    Low pass filter:Keep low & remove high frequencies

    High pass filter:Keep high & remove low frequencies

  • EE299 Lecture 723 Jan 2008

    More on FilteringCan also be used to “shape” frequency contentAny function of the form

    y(n)=a1 y(n-1)+a2 y(n-2)+…+ap y(n-p)+b0 x(n)+b1 x(n-1)+…+bm (n-m)

    can be used to shape/change frequency content (input is x(n), output is y(n))Leads us to the idea of generating signals

  • EE299 Lecture 723 Jan 2008

    Generating Signals: Electronic Music

    Notes(specify fundamental frequency)

    Cosine(+Harmonic)Generator

    FrequencyShaper

    Pick the filter with a frequency shape that matches the desired instrument

    instrument(frequency domain)

  • EE299 Lecture 723 Jan 2008

    Generating Signals: Speech

    Cosine(+Harmonic)Generator

    FrequencyShaper

    Pick the….• source signal depending on whether sound is voiced/unvoiced, and• filter with a frequency shape that matches the desired sound (ae,s,l,m,…)

    NoiseGenerator

    pitch

    switchvoiced

    unvoiced

    Alternative: glue together sounds snipped from a similar context

    (frequency domain)

  • EE299 Lecture 723 Jan 2008

    Changing a VoiceIdentify the system (frequency shape) associated with a particular soundUse a different source function with this system (e.g. robot-like voice with constant pitch)

  • EE299 Lecture 723 Jan 2008

    On to Images!Grayscale sampled images are 2D version of sampled soundsColor images: 3xgray (for mixing 3 colors)

    More on FridayImages have frequencies too!

    … so can have aliasing, filtering, etc.More next week

  • EE299 Lecture 723 Jan 2008

    Images vs. Sounds in Digital FormSounds:

    Bits = sec x samples/sec x bits/sampleSamples/sec = sampling rateBits/sample = quantizer resolution

    Gray-scale images:Bits = inches horizontal x pixels/in (H)

    x inches vertical x pixels/in (V)x bits/pixel

    Pixels/in = spatial resolutionBits/pixel = gray scale resolution

  • EE299 Lecture 723 Jan 2008

    Bits Mapping to Gray LevelsDecimal Binary Shade0 000 black1 001 very dark gray2 010 dark gray3 011 medium gray4 100 light gray5 101 very light gray6 110 hint of gray7 111 white

  • EE299 Lecture 723 Jan 2008

    Zooming in on an Image

    Try MATLAB zoom with imshow

  • EE299 Lecture 723 Jan 2008

    Effect of Changing Pixels/In

  • EE299 Lecture 723 Jan 2008

    Examples of Bit TradeoffsSee Orsak et al. pp 146-1471 bit per pixel (bpp) vs. 8 bpp rose

    Lecture 7:�Sounds & ImagesAnnouncementsGoals for TodayFiltering ReviewMore on FilteringGenerating Signals: Electronic MusicGenerating Signals: SpeechChanging a VoiceOn to Images!Images vs. Sounds in Digital FormBits Mapping to Gray LevelsZooming in on an ImageEffect of Changing Pixels/InExamples of Bit Tradeoffs