LIGHT & COLOR THEORY OVERVIEW - Radiant Vision Systems · o Not recommended, cannot guarantee...

Radiant Vision Systems | A Konica Minolta Company

LIGHT & COLOR THEORY OVERVIEW

Automated Visual InspectionLight & Color Global Support

3Radiant Vision Systems | A Konica Minolta Company

TODAY’S AGENDA

• Introduction to Light & Color• Instrumentation for

Photometry & Colorimetry • Characteristics of Imaging Sensors• Addressing Imaging Limitations

INTRODUCTION TO LIGHT & COLOR

WHAT IS LIGHT?

• Electromagnetic Radiation• EM Spectrum of light extends from UV to IR• Visible light consists of a very narrow band between UV and IR• Spectrum of light can be broken into discreet wavelengths that we

see as colors

LIGHT SOURCES AND SPECTRAL POWER DISTRIBUTIONS (SPD)

• Light emitted by a source can be defined by the source’s Spectral Power Distribution (SPD)

• SPD is the light output (power) at a given wavelength along the electromagnetic spectrum

• Different sources will have different SPDs

SPD OF COMMON LIGHT SOURCES

390 430 470 510 550 590 630 670 710

Wavelength λ (nm)

D65 (Sunlight)

Illuminant A

LCD Blue

LED Red

HeNe Laser

HUMAN VISION

The human eye has three types of cones (S, M, and L) that are each sensitive to a range of wavelengths of light.

A cone cannot detect the individual light wavelengths. Rather, a cone sees the sum total of light from all wavelengths under that cone’s spectral sensitivity curve.

350 450 550 650 750

2⁰ Cone Response

The human eye is sensitive to and

Wavelength λ (nm)

HUMAN PHOTOPIC VISION

SpectralSensitivity of

the Human Eye

The spectral limit of average human sight is:

380 – 830nm

Maximum luminous efficacy is:@ 555 nm (green)

SPD of Sunlight

PhotometricHuman Visual Perception

RadiometricAll Radiation

Total light outputLUMINOUS FLUX

lumens (lm)1 lm = 1 cd * 1 steradian

RADIANT FLUXWatts (W)

Light from a Direction

LUMINOUS INTENSITYcandela (cd)

1 cd = 1 lm/steradian

RADIANT INTENSITYW/sr

Light incident on a surface

ILLUMINANCE IRRADIANCEW/m2

lux (lx)1 lx = 1 lm/m2

foot-candle (fc)1 fc = 1 lm/ft2

Brightness

LUMINANCE RADIANCEW/sr * m2

1 cd/m2 = 1 nitfoot-lambert (fL)

1 fL = 1 cd/∏*ft2

PHOTOMETRIC VS. RADIOMETRIC UNITS

THE LANGUAGE OF LIGHT

Amount of light emitted in the range of a three-dimensional angular span

Candela (cd)

Luminous Intensity

Amount of light emitted from a light source or reflected back from a surface in a given direction

Candela per meter squared (cd/m2)

Luminance

Measure of the total luminous flux of a light source by integrating its intensity over its angular span

Lumen (lm)

Luminous Flux

The amount of light incident on a surface per unit area

Illuminance (lm/m2 or lux)

Illuminance

THE LANGUAGE OF LIGHT

THE FOUNDATIONS OF COLORIMETRY

In the mid-1800’s, James Clerk Maxwell demonstrated that mixing three primaries could produce the entire gamut of hues. He presented various colored samples to an observer, who then adjusted the brightness levels of two primaries shining on the target until the resulting white matched a white target.

ColorSample

WhiteTarget

W. Davis Wright and John Guild built upon this research in the late 1920’s to show how much red, green, or blue light energy was needed for a human observer to perceive any color across the visible spectrum.

CIE COLOR-MATCHING FUNCTIONS

Mathematical way to determine perceivedcolor and brightness from SPDs

CIE Color-Matching Functions

�𝒙𝒙�𝒚𝒚

�𝒛𝒛

CIE 1931 COLOR SPACE

CIE Color SpaceEvery visible color to the human eye.Colors defined by coordinates (Cx, Cy)

Color GamutProducible colors using a device’s primaries (a tristimulus color system via machine).• Using a linear combination of primary

colors (R,G, B) that occur at each corner of the triangle

GR B+ +

COLOR AND CIE TRISTIMULUS CURVES

• Three curves multiply the SPD, and the integrals are Tristimulus X, Y, and Z.

• Combinations of X, Y, and Z define color coordinates in two-dimensional space.

1931 CIE Color Space x = X/(X+Y+Z) y = Y/(X+Y+Z)

1960 CIE Color Space u = 4X/(X+15Y+3Z) v = 6Y/(X+15Y+3Z)

1976 CIE Color Space u’ = 4X/(X+15Y+3Z) v’ = 9Y/(X+15Y+3Z)

380 430 480 530 580 630 680 730 780

�𝒛𝒛Shade of Blue

CALCULATING CX & CY

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠Cx = XX+Y+Z

= Cy = YX+Y+Z

= 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠

SPD 1 (Blue LED)

𝐂𝐂𝐂𝐂 > 𝐂𝐂𝐂𝐂

CALCULATING CX & CY

380 430 480 530 580 630 680 730 780

�𝒛𝒛 Shade of Green

SPD 2 (Green LED)

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠Cx = XX+Y+Z

= Cy = YX+Y+Z

= 𝑠𝑠𝑠𝑠𝑠𝑠𝑙𝑙𝑠𝑠 𝐂𝐂𝐂𝐂 < 𝐂𝐂𝐂𝐂

CALCULATING CX & CY

380 430 480 530 580 630 680 730 780

�𝒛𝒛Shade of Red

SPD 3 (Red LED)

𝑠𝑠𝑠𝑠𝑠𝑠𝑙𝑙𝑠𝑠Cx = XX+Y+Z

= Cy = YX+Y+Z

= 𝑠𝑠𝑠𝑠𝑚𝑚𝑚𝑚𝑚𝑚𝑠𝑠 𝐂𝐂𝐂𝐂 > 𝐂𝐂𝐂𝐂

CALCULATING CX & CY

380 430 480 530 580 630 680 730 780

�𝒛𝒛

Mostly White

SPD 4 (Sunlight)

𝑠𝑠𝑠𝑠𝑚𝑚𝑚𝑚𝑚𝑚𝑠𝑠Cx = XX+Y+Z

= Cy = YX+Y+Z

= 𝑠𝑠𝑠𝑠𝑚𝑚𝑚𝑚𝑚𝑚𝑠𝑠 𝐂𝐂𝐂𝐂 = 𝐂𝐂𝐂𝐂

INSTRUMENTATION FOR PHOTOMETRY & COLORIMETRY

TECHNOLOGY COMPARISON

Light measurements systems include:• Illuminance Meters

• Luminance Meters

• Spot Colorimeters

• Spectroradiometers

• Imaging Colorimeters

TECHNOLOGY COMPARISON: SPOT SYSTEMS

TECHNOLOGY COMPARISON: IMAGING SYSTEMS

TECHNOLOGY COMPARISON

• No single technology is best for all measurement needs.• Imaging (sensor-based) instruments excel at:

• Contextual evaluation• Measuring uniformity• Identifying defects (pixels, blobs, artifacts)• Measuring multiple spots

(LED arrays)• Determining dimensions, distortion and focus quality• Performing advanced analysis

Imaging Photometers& ColorimetersScientific-grade camera systems with unique optical filters that replicate the human eye’s response to light.

CHARACTERISTICS OF IMAGING SENSORS

WHAT IS AN IMAGE SENSOR?

Representation ofSensor Pixels

One Sensor Pixel

WHAT IS AN IMAGE SENSOR?

• A rectangular matrix of photodetectors• Each detector measures incident flux

Detector matrix Optics form image on sensor

Each detector element (pixel) measures

incident flux

IMPORTANT IMAGE SENSOR CHARACTERISTICS

Resolution – The number of pixels in an image translates into spatial resolution, which determines the ability to distinguish fine detail within an image.

1MP Resolution 29MP Resolution

Pixel count – This translates into spatial resolution, which determines the ability to distinguish fine detail within an image.

THE IMPACT OF RESOLUTION

Higher-resolution imaging enables:

• Easier detection of contrast variations

• Pixel-level defect detection and correction (displays)

• Detection of subtle defects

Dynamic Range – The number of shades of gray. Larger dynamic range yields greater measurement precision and enables higher contrast ratio measurement.

Dynamic range:The ratio between the maximum possible signal and the “noise” level at the minimum signal of one measurement

• Measured in dB (decibels)

• Calculates a signal-to-noise ratio (SNR)

Signal

Background

Signal + Noise

Signal

Background

Signal + Noise

Physical Size – The size of the sensor’s active area determines the field-of-view when working with a given focal length lens.

Sensor

SensorLensLens

θ2θ1

IMAGING COLORIMETER CONFIGURATIONS

• 3-color commercial sensor• Monochrome, scientific-grade sensor with moving

filter wheel

3-COLOR SENSOR

• Sensor with color filters physically overlaid on each pixel.

• “Color” of each pixel is determined by using values of surrounding pixels.

3-COLOR SENSOR

• Advantages• Low Cost• No moving parts• Fast

• Disadvantages• Imperfect CIE match = Reduced color accuracy• Not 100% fill factor = Reduced effective

resolution• Small pixels = Low dynamic range• Low signal/noise ratio affects accuracy

• Sequential exposuresare made through CIEmatched color filters

• Neutral density filters modify the intensity ofeach wavelength

• Cooled, scientific-grade sensor is used

SCIENTIFIC-GRADE SENSOR WITH FILTER WHEEL

Imaging Lens

Color Filters

ND Filters

Sensor

TRISTIMULUS & NEUTRAL DENSITY FILTERS

Each color filter precisely matches a CIE tristimulus curve; which allows for NIST traceable results.

Color images are a combination of images taken through each filter.

ND Filter Wheel Bright Lights Modulated Lights Various filters

Greyscale Image Sensor Scientific Grade with Large Pixels Cooled to 5⁰C & Precision Calibrated Sensor with various resolution

Color Filter Wheel Precision made filters Xb Filter option Radiometric option

SCIENTIFIC-GRADE SENSOR WITH FILTER WHEEL

• Advantages• Best possible CIE match = highest color

accuracy• High dynamic range • Low noise• High quantum efficiency (for low light)• Higher sensor fill-factor

• Disadvantages• Higher cost• Slower measurement times• Moving parts (filter wheels)

Imaging Lens

Color Filters

ND Filters

Sensor

USAGE RECOMMENDATION

• 3-color commercial sensoro Not recommended, cannot guarantee color and luminance accuracyo Used only when gross relative color errors are needed (e.g., looking

for an orange LED instead of a red LED)

• Monochrome, scientific-grade sensor with moving filter wheel

• Recommended for R&D when accuracy and precision are important• Recommended for low-contrast display mura detection• Recommended for production applications where speed and

accuracy are both critical

ADDRESSING IMAGING LIMITATIONS

MEASUREMENT ERRORS

• Electronic• Optical• Mechanical

MEASUREMENT ERRORS

• Electronic• Photon (shot) noise• Dark (thermal) noise• Nonlinearity• Non-Uniform Pixel Response

• Optical• Mechanical

MEASUREMENT ERRORS

• Electronic• Optical

• Lens Distortion• Lens Cosine Fall-off• Lens Vignetting• Lens Spectral Dependency• Off-Axis (Keystone)

Distortion

• Screen Effects• CIE color filter mismatch• Filter wedge and

magnification

• Mechanical

MEASUREMENT ERRORS

• Electronic• Optical• Mechanical

• F-stop repeatability• Shutter repeatability

THE IMPACT OF NOISE

217Δ0O%

216Δ1

215Δ2

214Δ3

213Δ4

212Δ52.0%

211Δ6

Original Image

Small Amount of Noise Added

A Little More Noise Added

PHOTON (SHOT) NOISE

WellPixel Size

Photons per μm2

e-e-e-e- e- e-e-e-e-e- e- e-e-e-e-e- e- e-e-e-e-e- e- e-e-e-e-e- e- e-

Saturation(e-)

PIXEL SIZE / WELL DEPTH COMPARISONSensor Name Resolution Pixel Size

(μm)Electron Well

Depth (e-)

Sony ICX205 1392 x 1040 4.65 x 4.65 10,000

KAI-02050 1600 x 1200 5.5 x 5.5 20,000

KAI-08050 3296 x 2472 5.5 x 5.5 20,000

KAI-16050 4896 x 3264 5.5 x 5.5 20,000

KAF-1602E 1536 x 1024 9 x 9 100,000

KAF-6303E 3072 x 2048 9 x 9 100,000

KAF-0261E 512 x 512 20 x 20 200,000

KAF-1001E 1024 x 1024 24 x 24 240,000

-30 -20 -10 0 10 20 30 40

Temperature (C)

)DARK (THERMAL) NOISE

Dark Current vs Temperature

(logarithmic)Dark Current over Time

(linear)

0 1 2 3 4 5 6 7 8 9 10

CORRECTING DARK CURRENT

• Cool camera to reduce dark current• Limit exposure times of uncooled cameras to 10 sec• Capture a dark frame with the optical path blocked,

and then subtract it from subsequent measurements, pixel by pixel

• Average multiple images to reduce noise effects

CORRECTING FOR PIXEL RESPONSE AND LENS TRANSMISSION (LUMINANCE)

Uniform LuminanceLight Source

Camera

• Luminance flat fielding• Acquire image using

uniform light source

• Create correction matrix from acquired image• Multiply data images by correction matrix

Uncalibrated image of Uniform Light Source

Calibration image of Uniform Light Source

Corrected image

CORRECTING FOR PIXEL RESPONSE AND LENS TRANSMISSION (ILLUMINANCE)

• In addition to the problems described before, screens have the following problems:

Screen Blemishes Non-Uniform Reflectivity Keystone Distortion

• Illuminance flat fielding• Illuminate screen with

known light source and acquire exposure

Projection Screen

Known projectionlight source (AC-4001)

Camera

• Correct perspective distortion• Create correction matrix from acquired image• Multiply data images by correction matrix

Uncalibrated image of LCD projector with screen defect

Calibration image with calibration lamp

Corrected image

SENSOR LINEARITY CORRECTION AND Y-INTERCEPT ERROR

• Pixel response to light should be linear with time and amount of pixel well depth filled

• Sensitivity to light is not perfectly linear

• Map the sensor response v. saturation level• Use this map to create a correction table

CORRECTING LENS DISTORTION

• Software driven

Uncorrected image exhibiting barrel distortion in 7-mm lens.

The same image after a software lens distortion correction.

STRAY LIGHT CORRECTION

• Stray light affects measured Contrast Ratio

Without stray light correction; Luminance of black squares measured as 2-3 Nits

With stray light correction; Luminance of black squares measured as 0.6 Nits

MECHANICAL REPEATABILITY

• F-Stop Repeatability• Mechanical backlash in aperture springs

• Small error in aperture size leads to error in absolute luminance scaling.

• Always approach destination F-# from a lower F-#• Use electronically controlled lens – programmed to

eliminate this issue

• Shutter Repeatability• Minimum exposure limit driven by mechanical drive tolerances

• Can see an error in absolute luminance scaling or a luminance gradient over the image.

CIE COLOR MATCH

• Need to match each color channel response to corresponding CIE color curve in order to calibrate color properly.

EXAMPLE SENSOR SPECTRAL RESPONSE

CIE COLOR MATCH: 3-COLOR SENSOR• Response of color filters used for 3-color sensors

isn’t a good match to CIE curves.

CIE COLOR MATCH: COLOR FILTERS

• Excellent match to CIE color curves can be obtained by using individual color filters.

380 420 460 500 540 580 620 660 700

Imaging Lens

Color Filters

ND Filters

Sensor

CORRECTING COLOR FILTER RESPONSE MISMATCH

• One-color calibration• Scales each filter for the specific DUT

• Four-color calibration• Accurately measures all colors for the RGB device

• Multi-color calibration• Produces a compromise calibration for differing colors

• Single-filter brightness scaling• For brightness only, scales output to particular device

FILTER WEDGE AND MAGNIFICATION

• User sees a color shift between pixels in the image• Software driven correction• Applied to each filter

Error Category Source of Error Countermeasure

Sensor NoisePhoton noise (shot noise) Larger pixel size, deeper pixel wells, full use of available

electrons (75% saturation)

Dark noise (thermal noise) Sensor cooling, dark image subtraction

Sensor ErrorsPixel non-uniformity Flat-field calibrationResponse linearity Linearity testing, linearity correction tableResponse Y-intercept error Linearity correction table

Optical System Errors

Lens vignetting, optical system angular transmission differences Flat-field calibration

Lens distortion (short focal length lenses only) Lens distortion calibrationImage Off-Axis Distortion (keystoning) Geometric software correctionLight Scattering Stray light calibrationScreen defects (illuminance measurements only) Illuminance flat-field calibration

Camera ErrorsF-stop repeatability Approach from low F-stop, rescale, electronic lensShutter repeatability Minimum exposure time

Filter ErrorsCIE curve mismatch Color calibration, luminance scalingFilter wedge and magnification Color shift correction

OVERVIEW OF ERRORS AND COUNTERMEASURES

Radiant Vision Systems | A Konica Minolta CompanyRadiant Vision Systems | A Konica Minolta CompanyQuestions? Contact Info@RadiantVS.com

THANK YOU!

LIGHT & COLOR THEORY OVERVIEW - Radiant Vision Systems · o Not recommended, cannot guarantee...

Documents

Display Color Analyzer Model 7123 data sheet-1 · the problem of color analyzer measurement ... Y and Z filter compensation. The low luminance applied to the display or TV using the

Sensitivity to color variations Localization€¦ · Significance of Color CSFs for Vision ¾Part 2 Multiple Color Spatial Frequency Channels Luminance‐Color ¾Part 3 Introduction

arXiv:1606.05897v1 [cs.CV] 19 Jun 2016June 2016 (a) Content (b) Style (c) Output using [2] (d) Output with color preservation (color transfer) (e) Output with colour preservation (luminance

Chapter 7: Color perception Color is an important source of information independent of luminance (which we discussed extensively in Chapter 5). Color is

Seem - focalpointlights.com · luminance data ) ® ® luminance data )

ABSTRACT - doug.kerrdougkerr.net/Pumpkin/articles/HSL_HSV.pdfThe HSV and HSL Color Models Page 4 • Luminance. This tells us the “brightness” of the light whose color is being

CONSISTENT TONAL CORRECTION FOR MULTI-VIEW REMOTE …€¦ · Color Correction Luminance Compensation Tone Preprocess Original Images Corrected Images Resultant Images Structure Transferring

A New Disaster Prevention Product High-Luminance Glow-In ......Luminance of the Highest Level • High initial luminance that deteriorates comparatively slowly over time. • Excellent

Amplificateur de luminance

Modeling Human Color Perception under Extended Luminance … · 2012-10-23 · However, perceptual color appearance under extended luminance levels has not been studied, mainly due

Intelligent Luminance Detection System

Color Models & Color Tools in ImagesPlus - MLUnsold … · Saturation S = 0.0 is along the vertical axis and represents gray scale, Saturation increases as S ... luminance and color

2015 Luminance LED Catalog

FIELD LUMINANCE MAPPING OF THE SKYcltc.ucdavis.edu/sites/default/files/files/publication/201011-ies-field-luminaice... · FIELD LUMINANCE MAPPING OF THE SKY BACKGROUND Sky luminance

VA1921a Monitorul LCD - viewsonicglobal.com · resolution, black level, gamma curve, color and luminance uniformity, color rendition and image stability. • Product have been tested

Luminance War rooms · 2020-07-30 · Luminance War rooms This month, Luminance visited one of our clients at their office in the Netherlands to host one of our ‘War Rooms’. Luminance

Image Color Conversion by Illumination...of a white object under illumination S corresponds to the maximum luminance of the light source color with chromaticity coordinates of illumination

TD2220 LCD Display User Guide - ViewSonic · 2014-05-22 · TD2220 LCD Display User Guide Model No. VS14833 ... level, gamma curve, color and luminance uniformity, color rendition

Daylight luminance requirements for full-color, see

O B O A O A O B O B HEADER Single Color Print Ribbons · Four Color Paneled Print Ribbons Catalog # Color Ribbon Width Inch (mm) Color Panel Length Inch (mm) Overall Length Typical