Upload
mjunghwa
View
226
Download
0
Embed Size (px)
Citation preview
8/7/2019 Bitmapped Images-present
1/23
Bitmapped ImagesDigital Multimedia, 2nd edition
Nigel Chapman & Jenny ChapmanChapter 5
8/7/2019 Bitmapped Images-present
2/23
Bitmapped Images
Also known as r aste r g r aphicsRecord a value for every pixel in the image
Often created from an external source Scanner, digital camera,
Painting programs allow direct creation of
images with analogues of natural media,brushes,
8/7/2019 Bitmapped Images-present
3/23
8/7/2019 Bitmapped Images-present
4/23
Image Resolution
Array of pixels has pixel dimensions, but nophysical dimensions
By default, displayed size depends onresolution (dpi) of output device physical dimension = pixel dimension/ r esolution
Can store image r esolution (ppi) in image fileto maintain image's original size
Scale by device r esolution/image r esolution
8/7/2019 Bitmapped Images-present
5/23
Changing Resolution
If image resolution < output device resolution,must interpolate extra pixels
Always leads to loss of quality
If image resolution > output device resolution,must downsample (disca r d pixels)
Q uality will often be better than that of an image
at device resolution (uses more information) Image sampled at a higher resolution than that of intended output device is ove r sampled
8/7/2019 Bitmapped Images-present
6/23
Compression
Image files may be too big for networktransmission, even at low resolutionsUse more sophisticated data representation ordiscard information to reduce data sizeEffectiveness of compression will depend onactual image data
For any compression scheme, there willalways be some data for which 'compressed'version is actually bigger than the original
8/7/2019 Bitmapped Images-present
7/23
Lossless Compression
Always possible to decompress compresseddata and obtain an exact copy of the originaluncompressed data
Data is just more efficiently arranged, none isdiscarded
R un-length encoding ( RLE )
Huffmann codingDictiona r y-based schemes LZ77 , LZ78 , LZW (LZW used in GIF, licence fee charged)
8/7/2019 Bitmapped Images-present
8/23
JPEG Compression
Lossy technique, well suited to photog r aphs,images with fine detail and continuous tonesConsider image as a spatially varying signalthat can be analysed in the frequency domainExperimental fact: people do not pe r ceive theeffect of high f r equencies in images ve r y
accu r ately Hence, high frequency information can bediscarded without pe r ceptible loss of quality
8/7/2019 Bitmapped Images-present
9/23
DCT
Similar to Fourier Transform, analyses a signalinto its frequency components
Takes array of pixel values, produces an arrayof coefficients of frequency components in theimageComputationally expensive process timeproportional to square of number of pixels
Apply to 8x8 blocks of pixels
8/7/2019 Bitmapped Images-present
10/23
JPEG Q uantization
Applying DCT does not reduce data size Array of coefficients is same size as array of pixels
Allows information about high frequencycomponents to be identified and discardedUse fewer bits (distinguish fewer differentvalues) for higher frequency components
Number of levels for each frequencycoefficient may be specified separately in aquantization matrix
8/7/2019 Bitmapped Images-present
11/23
JPEG Encoding
After quantization, there will be many zerocoefficients
Use RLE on z ig-zag sequence (maximi zes runs)
Use Huffman coding of other coefficients (bestuse of available bits)
8/7/2019 Bitmapped Images-present
12/23
JPEG Decompression
Expand runs of zeros and decompressHuffman-encoded coefficients to reconstructarray of frequency coefficientsUse Inve r se Disc r ete Cosine T r ansfo r m to takedata back from frequency to spatial domainData discarded in quantization step of
compression procedure cannot be recoveredReconstructed image is an approximation(usually very good) to the original image
8/7/2019 Bitmapped Images-present
13/23
Compression Artefacts
If use low quality setting (i.e. coarserquantization), boundaries between 8x8 blocksbecome visibleIf image has sharp edges these becomeblurred
Rarely a problem with photographic images, butespecially bad with text
Better to use good lossless method with text orcomputer-generated images
8/7/2019 Bitmapped Images-present
14/23
Image Manipulation
Many useful operations described by analogywith darkroom techniques for altering photos
Correct deficiencies in image Remove 'red-eye', enhance contrast,
Create artificial effects Filters: stylize, distort,
Geometrical transformations Scale (change resolution), rotate,
8/7/2019 Bitmapped Images-present
15/23
Selection
No distinct objects (contrast vector graphics)Selection tools define an area of pixels
Draw selection (pen tool, lasso) Select regular shape (rectangular, elliptical, 1px
marquee tools) Select on basis of colour/edges (magic wand,
magnetic lasso)
Adjustments &c restricted to selected area
8/7/2019 Bitmapped Images-present
16/23
Masks
Area not selected is p r otected, as if masked by stencil
Can represent on/off mask as array of 1 bit perpixel (b/w image)Generalize to greyscale image(semitransparent mask) alpha channel
Feathered and anti-aliased selections Use as laye r mask to modify laye r compositing
8/7/2019 Bitmapped Images-present
17/23
Pixel Point Processing
Compute new value for pixel from its old value p' = f(p), f is a mapping function
In greyscale images, ppp alters brightness andcontrast Compensate for poor exposure, bad lighting, bring
out detail Use with mask to adjust parts of image
(e.g.shadows and highlights)
8/7/2019 Bitmapped Images-present
18/23
Adjustments in Photoshop
Brightness and contrast sliders Adjust slope and intercept of linear f
Levels dialogue Adjust endpoints by setting white and black levels Use image histog r am to choose values visually
Curves dialogue Interactively adjust shape of graph of f
8/7/2019 Bitmapped Images-present
19/23
Pixel Group Processing
Compute new value for pixel from its old valueand the values of surrounding pixels
F ilte r ing ope r ations
Compute weighted average of pixel values Array of weights k/a convolution mask Pixels used in convolution k/a convolution ke r nel
Computationally intensive process
8/7/2019 Bitmapped Images-present
20/23
Blurring
Classic simple blur Convolution mask with equal weights
Unnatural effect Gaussian blur
Convolution mask with coefficients falling off gradually(Gaussian bell curve)
More gentle, can set amount and radius
8/7/2019 Bitmapped Images-present
21/23
Sharpening
Low frequency filter 3x3 convolution mask coefficients all equal to -1,
except centre = 9 Produces harsh edges
Unsharp masking Copy image, apply Gaussian blur to copy, subtract
it from originalEnhances image features
8/7/2019 Bitmapped Images-present
22/23
Geometrical Transformations
Scaling, rotation, etc. Simple operations in vector graphics Requires each pixel to be transformed in
bitmapped image
Transformations may 'send pixels into gaps i.e. inte r polation is r equi r ed
Equivalent to reconstruction & resampling; tends todegrade image quality
8/7/2019 Bitmapped Images-present
23/23
Interpolation
N ea r est neighbou r Use value of pixel whose centre is closest in the
original image to real coordinates of ideal interpolated
pixelB ilinea r inte r polation
Use value of all four adjacent pixels, weighted byintersection with target pixel
B icubic inte r polation Use values of all four adjacent pixels, weighted using
cubic splines