Bettina Selig. Centre for Image Analysis. Swedish University of Agricultural Sciences Uppsala University

Transcription

1 Bettina Selig Centre for Image Analysis Swedish University of Agricultural Sciences Uppsala University

2 2 Electromagnetic Radiation Illumination - Reflection - Detection The Human Eye Digital Cameras Color spaces Pseudo coloring Multispectral data Chapter 6 in Gonzalez & Woods: Digital Image Processing, 3rd ed., 2008

3 3

4 4 Every imaging system selects one or several spectral windows. For gray scale image we have a single window, typically covering most of the visible spectrum.

5 5 The registered image depends on the spectral properties of: - light from, e.g., the sun or a lamp - light can be reflected, absorbed or transmitted - can be, e.g., a camera or the human eye

6 6 Achromatic light - White or uncolored light that contains all visual wavelengths in a complete mix. Chromatic light - Colored light. Monochromatic light - Light with a single wavelength, e.g., a laser. Colors we see are typically a mix of wavelengths. The dominant wavelength reflected by an object decides the color tone or hue. If many wavelengths are reflected in equal amounts, an object appears to be gray.

7 7, watts [W] The total amount of energy that flows from the light source., lumens [lm] A measure of the amount of energy an observer perceives. A subjective descriptor that is practically impossible to measure.

8 Additive Color Mixing 8 R + G + B = white e.g. printer e.g. screens

9 9 The is the region of the eye where you see sharp. Photoreceptor cells: are responsible for the perception of lightness and for color vision.

10 10 The Rods ( stavar) Rods are sensitive to light in the middle of the visual spectra, peak at 498 nm. There are about 100 million rod cells per eye. Slower response time than cones but about 100 times more sensitive to light. Night vision depends on the rods. Dominates the retina outside the fovea, this means that peripheral vision depends on the rods.

11 11 The Cones ( tappar) About 6 million cone cells per eye. Most localized in the fovea. Three types of cones: (blue), peak at 420 nm, (2%) (green), peak at 534 nm, (33%) (red), peak at 564 nm, (65%) Dotted line - rods S - cones for short wavelengths M - cones for medium wavelengths L - cones for long wavelengths Normalized sensitivity of rods and cones

12 12 12 types of color receptors extending the view into the ultraviolet. Compound eyes with three different regions giving each eye trinocular vision. They are very colorful not only in our visual spectra.

13 13 Based on a large scale study with volunteers evaluate color samples. Representing all of the colors visible to the average person. A 2D projection of the XYZ color space is the CIE chromaticity diagram (to the right). Standard white when X=Y=Z CIE 1931 xy chromaticity diagram

14 14 The spectral measurement process The observed color can be defined in the CIE XYZ color space via calculation of the tristimulus values X, Y, and Z as follows:

15 15 Much like the human eye a digital camera has sensors sensitive to three colors. In a CCD chip they are differentiated by a Bayer filter pattern. Values are then interpolated to a full RGB image. Bayer mosaic Camera

16 16 Early color spaces found in the history of art: proposed this one in A color space based on three elementary colors, red, yellow and blue.

17 17 RGB/CMY, Red Green Blue / Cyan Magenta Yellow RGB is closer to the physiological side of our vision (the three cone types) rather than the psychological. [C M Y] = [1 1 1] - [R G B]

18 18 HSV/HSL Hue, angle Saturation, radius ( ) (Färgton) (Mättnad) Value, Lightness, height (Ljushet) HSV The HSV and HSL color spaces has intensity decoupled from color information. Example for HSV color picker HSL

19 19 Difference between HSV/HSL The difference is how saturation and lightness/value hang together. For some people it is non intuitive to be have almost white colors that are fully saturated while others likes the lightness property in HSL.

20 20 NCS Natural Color System Designed to have a notation more intuitive and readable than RGB, HSL and alike. Used by designers, architectures and in marketing. Color are described in relation to six elementary colors: black, white, red, yellow, green and blue. The yellow in the Swedish flag is defined in NCS as: 0580-Y10R 5 % darkness, 80 % saturation, yellow with 10 % red.

21 21 CIE L*a*b* or CIELAB The most complete color space specified by the International Commission on Illumination, CIE in Created to: represent all colours visible to the human eye. serve as a device independent model to be used as a reference. be perceptually uniform - equal distance should have equal perceptual difference. Remember the 200 years old color model:

22 22 Even more color spaces is similar to CIE L*a*b* Used in the JEPG file format and for digital TV. Uses the fact that the human eye is more sensitive to variation in lightness than in hue and saturation. is similar to YCbCr, used for analogue TV. is the CMY colour space with a black component added, used for printing where a black pigment is used along with Cyan, Magenta and Yellow.

23 23

24 24 Gaussian noise in all color channels In a HSL representation the noise is most apparent in the H and S channel.

25 25 In general all image processing methods used for grey level images can be used for color images. They can be carried out on each of the color channels or for example on the intensity only. There is no right or wrong, but the result differ. Use the intensity to find edges between areas and color values as similarity measure.

26 26 You might end up with color artifacts if the H-channel is filtered. Remember that the Hue channel is in degrees 0 0 to (both red). Gaussian filtered hue channel

27 27 Histogram equalization on all channels in HSV can make things strange. Only used on the V channel the expected result is acquired. Original Histogram equalization in all channels Histogram equalization in V channels

28 Segmented part of hue shown on the value channel. 28 Using a intensity decoupled color space segmentation based on color can be relative intuitive. Setting an interval for the hue around the hue value for red in HSV space the red part of the fish is segmented.

29 29 A color space can be close to the hardware or close to the application. RGB is close to the output from a CCD, etc. Decoupled intensity can be very useful in image processing making it possible to use many gray level methods intuitively. Some spaces like HSV has a difficult transformation from, e.g., RGB. Singularities may exist (e.g., hue for gray, saturation for black). Regardless of which color space is used RGB is often the color space for the displaying device.

30 30 The CIE chromaticity diagram contain all visible fully saturated colors. Standard white when X=Y=Z The typical monitor gamut with three primary colors cannot produce all colors. A color printer with five pigments produce another gamut. CIE 1931 xy chromaticity diagram with a typical monitor gamut and a printer gamut is a subset of colors which can be accurately represented in a given circumstance.

31 31 For visualization The eye can distinguish between ~30 different gray levels and about different colors. A gray level image can be displayed as a color image for easier visual inspection. Note: the information content is not changed. This is often used in modalities when the samples does not correspond to the visual spectrum.

32 32 PET FDG tracer is used, high intensity related to high brain activity.

33 33 PET + CT

34 34 Why restrict our imaging system to the borders of human vision? Multi spectral data is common in remote sensing, astronomy, thermography, etc. Multispectral satellite data with 7 spectral bands of the Uppsala urban area.

35 35 Every imaging system selects one or several spectral windows. Hence, the spectral dimension is unavoidable. There are several ways to represent the color information in the visible spectra. (Color spaces) The RGB color space is usually unavoidable since viewing devices (LCD/CRT screens) use it. Choose color space based on the specific application and situation.