The photographic Camera
Transcription
The photographic Camera
The photographic Camera many pictures from Lana Lazebnik, Fredo Durand, Steve Seitz, Alyosha Efros Imaging • How to record light reflected from an object? • First attempt • place a light-sensitive plane in front of the object • WHat will the resulting image look like? Imaging • How to record light reflected from an object? • First attempt • place a light-sensitive plane in front of the object • WHat will the resulting image look like? Imaging • How to record light reflected from an object? • First attempt • place a light-sensitive plane in front of the object • WHat will the resulting image look like? Imaging • How to record light reflected from an object? • First attempt • place a light-sensitive plane in front of the object • WHat will the resulting image look like? Imaging • How to record light reflected from an object? • Second attempt • Points shall be mapped to points → only one ray for each point • place an aperture between object and image plane Imaging • How to record light reflected from an object? • Second attempt • Points shall be mapped to points → only one ray for each point • place an aperture between object and image plane Imaging • How to record light reflected from an object? • Second attempt • Points shall be mapped to points → only one ray for each point • place an aperture between object and image plane Imaging • How to record light reflected from an object? • Second attempt • Points shall be mapped to points → only one ray for each point • place an aperture between object and image plane Camera Obscura • Principle known for a long time • in China: Mo-Ti (470-390 BCE) • in Europe: Aristoteles (384-322 BCE) • since late middle ages used as drawing aid and scientific instrument • Roger Bacon (1214-1294) • Leonardo da Vinci (1452-1519) • Johannes Kepler (1571-1630) Gemma Frisius, 1558 Influence of aperture http://www.debevec.org/Pinhole/ Influence of aperture • How big should the hole be? • • small opening → small cone of rays → crisp image limit: too little energy collected, diffraction Influence of aperture Use of a lens • Light rays are deflected • Rays from one object point meet in one image point Use of a lens • Light rays are deflected • Rays from one object point meet in one image point Use of a lens focal point f • Light rays are deflected • Rays from one object point meet in one image point • parallel rays meet at the focal point Use of a lens “blur circle” • Light rays are deflected • distance where rays meet depends on distance object-lens • only for a specific distance image is “in focus” Lens equation • Geometric optics • • Assumption: thin lens from equal triangles y� D� = y D D’ y� D� − f = y f D� D� − f = D f D f y y’ Lens equation • Geometric optics • points which fulfill the equation are in focus D� D� − f = D f D’ f | 1 · � D D 1 1 1 + = D� D f Depth of field http://www.cambridgeincolour.com/tutorials/depth-of-field.htm Depth of field The depth of field depends on the aperture • small aperture → larger depth of field • but: less light → longer exposure needed Depth of field Aperture angle Aperture angle Aperture angle f f • Angle depends on • • • focal length f size of light sensor d d φ = tan = tan φ 2f 2f shorter focal length → larger aperture angle Aperture angle Aperture angle wide-angle, short focal length object close to camera narrow-angle, long focal length object far away Focal length • Perspective distortion depends on focal length Focal length • Perspective distortion depends on focal length wide-angle standard angle tele lens Lens errors • Radial distortion • in real lens systems magnification depends on the distance from the optical axis no distortion pin cushion barrel Lens errors • Chromatic aberration • • Refraction index of lens depends on wavelength no “correct” focal length for different colours image center image border Lens errors • Sphärische Aberration • ideal lens is infinitely thin • real lens: focus point is closer for rays from the border Lens errors • Vignetting • with growing distance from optical axis more and more light is lost • brightness decreases towards the image border Digital Sensors • Sensor is a raster of tiny photodiodes • individual diodes convert incoming photons to electrons • electric charge generates voltage • two types – Charge coupled device (CCD) – Complementary metal oxide semiconductor (CMOS) Sensor Technology • CCD (charge coupled device) • charge genenrated by diode is moved across the chip and read out in teh corner (i.e. the resulting voltage is measured) • CMOS (complementary metal-oxide semi-conductor) • Charge is converted directly at the pixel, amplified and read out • Digitisation • in both cases the measured voltage is converted to a discrete birghtness value with an A/D coverter Colour • Bayer Filter • • missing observations in each channel are interpolated (demosaicing) more green pixels, because human eye is most sensitive at ca. 550nm Sensitivity of visual system Colour Moiré Colour Moiré • Sampling Problem • Demosaicing assumes that spatial structures have lower frequency that pixel raster • otherwise same colours repeatedly bright/dark → Intensity pattern miss-interpreted as colour pattern Three-chip Cameras • Colour components split with prisms • three separate sensors • every channel has a measurement at every raster position • no interpolation artifacts, better image quality CCD(R) CCD(G) CCD(B) Separate Cameras • In aerial mapping cameras • multiple “camera heads” - panchromatic (high resolution) R,G,B,IR (lower resolution) • multiple line sensors green (29 Mpix) blue (29 Mpix) pan (260 MPix) red (29 Mpix) infrared (29 Mpix) Aerial cameras • digital only since a few years • Recording was the last step to go digital • before: scanning of analog film • reason were technical limits of building sensors Aerial cameras • digital only since a few years • Recording was the last step to go digital • before: scanning of analog film • reason were technical limits of building sensors Aerial cameras • Advantages of digital recording • • • • Lower cost: film, development, scanning not needed no extra cost for more images → higher accuracy better radiometric quality → higher accuracy more spectral channels: no limitations due to film Analog, GSD = 20cm Digital, GSD = 20cm (Leica ADS 40) Aerial cameras • More images • no extra cost for more images → higher accuracy - fewer occlusions - larger redundancy Aerial cameras • More images • no extra cost for more images → higher accuracy - fewer occlusions - larger redundancy Aerial cameras • Radiometric quality • better radiometric quality → higher accuracy - more precise localisation - better image interpretation Film GSD=0.5m R=265µm Ultracam GSD=0.4m R=145µm Aerial cameras • Radiometric quality • Resolution >12 bit (4096 brightness values) Aerial cameras • More spectral channels • Film has only three layers (RGB, CIR) • in digital devices a separate camera per colour • Colour has lower resolution → pan-sharpening panchromatic RGB pan-sharpened CIR pan-sharpened Aerial cameras • Pan-sharpening • Fusion of high-resolution panchromatic image and lowerresolution colour image Aerial cameras • Spectral saensitivity • Channels delimited more clearly in digital technology • Spatial resolution of colour channels significantly lower • • pan-sharpening can lead to colour bleeding at borders exception: line cameras Aerial cameras - Examples • Microsoft (Vexcel) Ultracam Eagle • • • High-resolution image stitched together from 9 tiles 262 MPix panchromatic additional cameras for R, G, B, IR Aerial cameras - Examples • Microsoft (Vexcel) Ultracam Eagle “master-cone” defines image geometry Recording time synchronised to create image Tiles are registered by digital image processing two different focal lengths: 80 / 210 mm flying direction 20 010 Pixel (x) 13 800 Pixel (y) • • • • Aerial cameras - Examples • Microsoft (Vexcel) Ultracam Eagle Aerial cameras - Examples • Zeiss/Intergraph DMC II • • • monolithic CCD-chip 235 / 250 MPix focal length 112 mm Aerial cameras - Examples • Analog cameras • Zeiss RMK-Top • Leica RC30 Aerial cameras - Examples • Leica ADS 80 three-line camera • • • • • • 12’000 pixels / line scan width (swath angle) 64º 3x panchromatic, viewing directions -16º, 0º, 40º nadir has 2 staggered lines for improved resolution (“super-resolution”) 2x R, G, B, NIR radiometric resolution 12 bit Aerial cameras - Examples • Medium format cameras • ca. 40-60 MPix • several vendors • applications: lower accuracy/resolution, smaller projects, UAVs, combination with LiDAR, ... IGI DigiCAM Z/I RMK D Aerial cameras - accessories • GNSS/IMU unit • Navigation, automatic camera trigger • Observation of orientation parameters • Accuracy ca. 3x lower than indirect sensor orientation Applanix POSAV Aerial cameras - accessories • Integrated system • Applanix DSS • application: “out-of-the-box” orthophoto generation with moderate accuracy (direct geo-referencing)