Download presentation
Presentation is loading. Please wait.
Published byElmer Merritt Modified over 8 years ago
1
Tytus Bernas Biological applications of image processing
2
JC Russ (1998) Image processing handbook Image acquisition system
3
JC Russ (1998) Image processing handbook Digitization…
4
Coordinate system of digital image (square pixels)... and its effects
5
Pixels T
6
Advantage of (mega)pixels
7
Microscope resolution
8
d ~ R Airy d = 0.6100 λ /NA d ~ JDD d = 0.4750 λ /NA d ~ FWHM d = 0.5015 λ /NA Microscope resolution
9
frequency 1x 2x 3x Nyquist criterion theoretical: 2x practical: 3x Perfect sampling
10
Fourier transform PSF MTF
11
http://www.microscopyu.com/ Modulation transfer function (MTF)
12
Widefield: Δx = λ em / (4 n sin α) Δz = λ em / (2 n (1- cos α)) Confocal: Δx = λ ex / (8 n sin α) Δz = λ ex / (4 n (1- cos α)) PSF MTF Perfect sampling
13
Aliasing
14
E.H.K. Stelzer et al. (1998) J. Microscopy 189 p 15- 24 PSF/MTF and noise
15
JC Russ (1998) Image processing handbook http://micro.magnet.fsu.edu/primer/ Camera: an array detector
16
Charge Coupled Device JC Russ (1998) Image processing handbook, http://micro.magnet.fsu.edu/primer/ CCD readout
17
http://www.astro.uu.se/~johwar/AMAST/sxvh9.html NOISE ! readout dark quantum blooming SNR = PQ e t / [PQ e t + Dt + Nr 2 ] 1/2 linearity Problems of (digital) cameramen
18
http://www.diracdelta.co.uk/ Quantization accuracy gain & ofset binning (pixel size) acquisition time Qantization, detector settings
19
low offset high offsethigh gain low gain Gain and offset
20
Image is a two-dimensional function of coordinates: I=f(x,y) Minimum: 0 (or 1) Maximum: dynamic range Computer image = table of numbers
21
Dynamic range
22
1, 2, 3, 4, 5, 6, 7, 8, 9...er aha! 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 20, 21..... 98, 99 100, 101..... and so on 1234 1 1000 1 10 3 2 100 2 10 2 3 10 1 10 1 4 1 4 10 0 00 11 210 311 4100 5101 6110 7111 81000 91001 101010 111011 121100 131101 141110 151111 10101 0802308023 1412214122 0202102021 1112011120 1 16 1 2 4 16 + 0 + 4 + 0 +1 = 21 Binary notation
23
product (AND) pq r 11 1 10 0 01 0 10 0 sum (OR) pq r 11 1 10 1 01 1 10 0 negation (NOT) p r 0 1 1 0 AND 1001 1100 1000 OR 1001 1100 1101 NOT 1001 0110 ABA AND B A OR BA exOR BNOT A Logic
24
Segmentation with binary mask I > t => 1 I 1 binarization log. AND
25
Mapping function (LUT)
26
image dark <> bright contrast low <> high Whole image: histogram
27
"What is truth?" Pontius Pilate (John 18:38) Conversion
28
Image improvement – convolution (averaging) 111 111 111 3x3 (1) 3x3 (3) 7x7 (1) 1111111 1111111 1111111 1111111 1111111 1111111 1111111 3x3 pix. 7x7 pix.
29
Better than average - median 111 111 111 3x3 median 3x3 gauss.
30
Enhancement of details - convolution 000 10 000 8 3x3 laplacian 3x3 derivative 9 3x3 hi-pass
31
Additive space: Red Green Blue Subtractive space : Cyan Magenta Yellow Hue Saturation Brigtthess Lightness a b Colour spaces
32
Warm and cold colours
33
Colour components I
34
Colour components II
35
(Pseudo)colours
36
Where to store it File typeBit depthColor coding Transp. (bits) compressionPortabilityRemarks losslesslossy BMP1/4/8grayscale ---- RLE ---- Poor (Microsoft format) Inefficient compression 1 24RGB- GIF8RGB color table1LZW-ExcellentNo grayscale support. JPEG24YUV--JPEGGood Very efficient compression 1 (with distorion). No RGB and grayscale support. JPEG20001/4/8color table38 2 JPEG200 fair Very efficient compression 1. Relatively new format intended to replace jpeg. 8/16grayscale 24/48RGB 36/64CMYK PNG1/2/4/8 grayscale or RGB color table 8deflation-good Efficient compression The format developed to replace GIF in www. 24/48RGB TIFF1/4/8RGB color table8 Deflation (ZIP), RLE, Huffman, LZW JPEGgood Format offers large array of features, which may be extended further using specific tags. Only limited number of these is supported universally, though. Efficient compression 1. 1/4/8/16grayscale 24/48RGB, YUV 32/64CMYK
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.