Transcription

1 HDR

2 Čo s tým ďalej?

3

4

5 Brightside DR37-P Dolby Over 3,000 cd/m 2 brightness cd/m 2 black level Contrast ratio > 200,000:1 High-definition 1,920 x 1, inch screen 16 bits per color príkon ~ 1600W hmotnosť ~ 70kg cena ~ USD

6 Technology Resolution Display size 47 Panel aspect ratio 16:9 Number of real colours Number of LED 2202 SIM2 HDR47E S 4K Dolby HDR LCD Display with individually controlled LED backlight modulation 1920 x 1080 pixels 16 bit per channel Brightness 4000 cd/m 2 ANSI Contrast >20.000:1 FULL ON/OFF Contrast White point virtually infinite (> :1) 6500K (totally adjustable 5000k 9000k) LED B.L.U. life time hours

7

8 Tone Mapping 10-6 High dynamic range to 255

9 Tone Mapping

10 Typy operátorov Globálne Rovnaká nelineárna krivka aplikovaná na všetky body obrazu Lokálne Adaptačná krivka je rôzna pre každý bod podľa jeho okolia Frekvenčné DR je menený podľa frekvencií obsiahnutých v obraze Gradientné Mení sa derivácia obrazu

11 A review of tone reproduction techniques Kate Devlin DCS, University of Bristol, 2002

12 Globálny Operátor A Contrast-based Scalefactor for Luminance Display Ward, 1994 Reinhard, 2002 Adaptive Logaritmic Mapping For Displaying High Contrast Scenes. Drago, Myszkowski, Annen, Chiba, 2003

13

14 Photographic Tone Reproduction for Digital Images Reinhard et al Zonálny systém Ansela Adamsa Kľúč: subjektívna miera svetlosti obrazu

15

16 Klúč scény - log priemer: L w 1 N exp x, y log L w x, y Globálny lineárny operátor: L a L x, y L x, y w w a klúč výslednej scény používa sa 0.18, ale môže byť aj 0.09, 0.045, 0.36 alebo 0.72

17

18 Globálny nelineárny operátor Globálny nelineárny operátor s parametrom y x L y x L y x L w w d, 1,, y x L L y x L y x L y x L w white w w d, 1, 1,, 2

19 Vylepšenie kontrastu pre LDR

20 Pre HDR niekedy nedostatočné

21 Dodging and Burning centrum-okolie funkcia s y x V s a s y x V s y x V s y x V,, / 2,,,,,, Určíme optimálnu škálu s m

22

23 Lokálny operátor D & B L d x, y 1V 1 Lw x, x, y y, s m x, y Korekcia farby C d L L d w C w

24

25 Gradient Domain HDR Compression Fattal et al., 2002 Jednoduchá myšlienka Identifikácia veľkého gradientu v rôznych škálach Progresívne zoslabenie veľkosti gradientu Rekonštrukcia obrazu z nového gradientu

26

27 Mapa potlačenia Gaussovská pyramída Škálové funkcie potlačenia log I k ( x, y) H k ( x, y)

28 Konštrukcia výslednej funkcie

29

30 Fast Bilateral Filtering for the Display of HDR Images Frédo Durand & Julie Dorsey Vstupné HDR Fast Bilateral Filtering [Tomasi a Manduchi 1998] výsledok Intenzita Báza Redukcia kontrastu Báza Rýchly bilaterálny filter Detail zachováme Detail Farba Detail = Intenzita báza Farba

31

32

33 Perception Motivated Hybrid Approach to Tone Mapping Martin Čadík, 2007

34 Enhancement map

35 Time-dependent visual adaptation for realistic image display Pattanaik, Tumblin, Yee, Greenberg, 2000 multi-scale observer operator attempts to model all steps within the human visual system currently known well enough to be modelled

36

37 Adaptive Logarithmic Mapping for Displaying High Contrast Scenes F. Drago, K. Myszkowski, T. Annen, and N. Chiba, 2003 Time-Dependent Visual Adaptation for Realistic Image Display S.N. Pattanaik, J. Tumblin, H. Yee, and D.P. Greenberg, 2000 Dynamic Range Reduction Inspired by Photoreceptor Physiology E. Reinhard and K. Devlin, 2004 Photographic Tone Reproduction for Digital Images E. Reinhard, M. Stark, P. Shirley, and J. Ferwerda, 2002 Fast Bilateral Filtering for the Display of High-Dynamic-Range Images F. Durand and J. Dorsey, 2002 A Tone Mapping Algorithm for High Contrast Images M. Ashikhmin, 2002 Gradient Domain High Dynamic Range Compression R. Fattal, D. Lischinski, and M. Werman, 2002 A Perceptual Framework for Contrast Processing of HDR Images R. Mantiuk, K. Myszkowski, and H.-P. Seidel, 2006

38 Porovnanie operátorov

39 tone mapping: upravenie intenzity gamut mapping: upravenie farieb

40 Gamut = rozsah farieb, ktoré dokáže zariadenie zobraziť, alebo rozsah farieb v obraze Mapovanie gamutu: Napr. Prevod farieb obrazu do farieb zobraziteľných tlačiarňou

41

42 Gamut CRT monitora (drôtený model) Gamut rôznych tlačiarní (pevné teleso) CIELAB space

43

44

45 Deskriptor hranice gamutu

46

47

48 Gamut mapping algoritmy globálne GMA Kompresia gamutu Orezanie gamutu lokálne GMA Závisí od okolitých bodov Zachováva detaily

49 Gamut mapping algoritmy Obrazovo nezávislé Menej časovo náročné Horší výsledok Obrazovo nezávislé Časovo náročné Lepší výsledok

50 Typy mapovania Orezanie gamutu Mení body len mimo gamutu Kompresia gamutu Mení body aj vnútri

51 Len červná farba Celý gamut

52 Orezanie Ortogonálne Radiálne L+C Horizontálne (C) Vertikálne (L)

53 Kompresia Lineárna kompresia L Lineárna kompresia C Nelinárna kompresia C

54 Radiálna kompresia L+C

55

56 CBIR content-based image retrieval

57 Problems with Image Retrieval A picture is worth a thousand words The meaning of an image is highly individual and subjective

58 What is the topic of this image? What are right keywords to index this image? What words would you use to retrieve this image?

59 CBIR Art Collections e.g. Fine Arts Museum SF, Hermitage Medical Image Databases CT, MRI, Ultrasound, The Visible Human Scientific Databases e.g. Earth Sciences General Image Collections Corbis, Getty Images The World Wide Web

60 Two Classes of CBIR Narrow vs. Broad Domain Narrow - špecifický Medical Imagery Retrieval Finger Print Retrieval Satellite Imagery Retrieval Broad - všeobecný Photo Collections Internet

61 Challenges Semantic gap The semantic gap is the lack of coincidence between the information that one can extract from the visual data and the interpretation that the same data have for a user in a given situation. User seeks semantic similarity, but the database can only provide similarity by data processing. Huge amount of objects to search among. Incomplete query specification. Incomplete image description.

62 What is a query? an image you already have a rough sketch you draw a symbolic description of what you want e.g. an image of a woman with a child

63 Aims of an user in CBIR browsing through a large set of images from unspecified sources category search retrieve arbitrary image representative of some class target search - search for a precise copy of an image (copyright protection) search for a picture to go with a broad story or to illustrate a document

64 On the right a block of text from Moby Dick. This text is processed to obtain nouns, verbs, adjectives and adverbs and the terms are disambiguated by a voting process. The resulting text is used as a query to Barnard et al. s joint probability model, where the search returns images that have high joint probability with the collection of words. On the left, the images returned by this query. The query appears to be very successful (among other things, there s a picture of a whaleboat with sailors in it harpooning a whale).

65 What is CBIR? Query Image Retrieved Images Image Database Feature Space Image Features Similarity Assessment