Digitization and fundamental techniques

Transcription

1 Digitization and fundamental techniques Chapter Robin Strand Centre for Image analysis Swedish University of Agricultural Sciences Uppsala University

2 Outline Imaging Digitization Sampling Labeling Interpolation Connectivity

3 Digitization What is a digital image? How are images acquired? How are resolution, sampling and quantization related?

4 Digitization

5 Electromagnetic waves

6 Signals for digital imaging Electromagnetic imaging Radio range, medicine, MRI Astronomy, astronomical radiation MRI image of knee

7 Signals for digital imaging Electromagnetic imaging Microwave range, RADAR Independent of weather and illumination conditions Reflected energy recorded by an antenna

8 Signals for digital imaging Electromagnetic imaging Visible range, Standard camera, light microscopy Automated license plate reading Light microscopy

9 Signals for digital imaging Electromagnetic imaging X-ray range: X-Ray Imaging, Medical diagnostics, industry: CT, micro-ct Tomography Chest X-ray

10 Signals for digital imaging Electromagnetic imaging Gamma range: Gamma-Ray Imaging, nuclear medicine (injected radioactive isotope), PET astronomical observations PET image

11 Signals for digital imaging Other, Scanning electron microscopy (SEM) Transmission electron microscopy (TEM) Ultrasound

12 Characteristics of wavelengths in visible and infrared range NASA's LANDSAT satellite Name Wavelength Characteristics Visible blue µm Maximum water penetration Visible green µm Plant vigor Visible red µm Vegetation discrimination Near infrared µm Biomass and shoreline mapping Middle infrared µm Moisture content of soil and vegetation Thermal infrared µm Soil moisture; thermal mapping Middle infrared Mineral mapping µm

13 Sensors To image something we need sensors that can capture the energy of the signal. The output from each sensor element is a positive finite value proportional to the received energy. Sensor array Bayer filter

14 Digitization The continuous value needs to be possible to represent in a computer, it needs to be digitized Gray level quantization - Discretization of amplitude values Spatial sampling - Discretizing a continuous function in terms of coordinate value Uniform sampling (sampling on a lattice) Non-uniform sampling (adaptive, denser sampling where there is much details)

15 Digitization

16 Digitization

17 Interpolation Known data is used to estimate values at positions where the data is unknown.

18 Nearest neighbor interpolation The pixel (x,y) is assigned the grey level of the input pixel closest to (x,y)

19 Bi-linear interpolation The grey level assigned to (x,y ) is a combination of the 4 neighboring values of (x,y) First in x direction: Top=(1 Δx)V1+ ΔxV3 Bottom= (1 Δx)V2+ ΔxV4 Then in y direction: G=(1 Δy)Top+ ΔyBottom

20 Interpolation Nearest neighbor bi-linear

21 Spatial sampling

22 Gray level quantization

23 Common quantization levels Image values when using integers, in interval [0, 2n 1]. Bits n=1 n=5 n=8 n = 16 n = 24 Interval [0, 1] [0, 31] [0, 255] [0, 65535] [0, ] Comment binary image what the human can resolve locally 1 byte, very common common in imaging systems common for color images (3 8 bit for RGB)

24 Aliasing If sampling frequency is too low, all details will not be captured. New patterns can appear!

25 Aliasing This image was scanned from a magazine, resulting in a pattern due to the frequency of the raster in the printing.

26 Choice of sampling What will the image be used for? What are the limitations in memory and speed? Will we only use the image for visual interpretation or do we want to do any image analysis? What information is relevant for the analysis (i.e., color, spatial and/or graylevel resolution)?

27 Image resolution higher resolution -> more image details Pixel resolution # picture elements / area (pixels per inch) Spatial resolution how closely lines can be resolved in an image - sensor size, lens quality, organization of the pixels,... Temporal resolution time of exposure number of images per second Spectral resolution range of wave-lengths number of colors Gray scale resolution (quantization) radiometric - measurement of electromagnetic radiation how many gray levels

28 Connectivities 4-connectedness objects connected through edge-neighbors 8-connectedness objects connected through edge- and vertex-neighbors

29 Labeling

30 Labeling 4-connected 8-connected

31 Color images f (x,y,b) x,y spatial parameters b spectral

32 Volume images Volume: f (x,y,z)

33 Acquiring volume images Tomography The word was derived from the Greek word tomos which means "part" or "section", representing the idea of "a section", "a slice" or "a cutting". Definition according to Merriam-Webster: A method of producing a three-dimensional image of the internal structures of a solid object (as the human body or the earth) by the observation and recording of the differences in the effects on the passage of waves of energy impinging on those structure

34 Tomography, CT

35 Tomography

36 Summary Digitization Resolution Don't oversample! What information is needed? Don't undersample! Aliasing. Interpolation Labeling Volume, color images