CSI: Rombalds Moor Photogrammetry Photography

Transcription

1 Photogrammetry Photography

2 Photogrammetry Training 26 th March 10:00 Welcome Presentation image capture Practice 12:30 13:15 Lunch More practice 16:00 (ish) Finish or earlier

3 What is photogrammetry 'photo' - light 'gram' - drawing 'metry' - measurement The art and science of obtaining reliable measurement by means of images (Remote Sensing and Photogrammetry Society, 2004) Photogrammetry is underpinned by the principle of triangulation. If imagery of the same subject is taken from two different locations, lines of sight, or rays can be developed from each camera to points visible on both images. A 3 dimensional x,y,z coordinate can then be calculated from the intersection.

4 Photogrammetry Today Photogrammetry is an extremely flexible tool and is used in many different disciplines / fields such as: Quantity surveying Planning / surveying Topographic mapping Architecture Engineering CSI (ballistics and surveying road accidents) Geological measurement Heritage

5 CSI Photogrammetry Equipment A digital camera Scales Software Tripod Parasol Control points / targets 3D monitor / Glasses

6 Metric Camera Photogrammetric photography is typically acquired using cameras with fixed focus lens often referred to as metric cameras for which the precise focal length of the lens and its distortion characteristics are known. This ensures that the same focal length, and hence image geometry, is used for each image acquired resulting in high accuracy processing.

7 Digital Camera Lower cost digital cameras (compact and DSLR) can also be used for photogrammetry providing their image geometry is known the process of determining the geometry is referred to as calibration Focal Length e.g. 24mm (CSI lens) Lens Distortion Principle Point either of two points on the principal axis of a lens where the object and image have the same size and are not inverted in relation to each other Resolution sensor / pixel size CSI using 2x Nikon D Megapixels (4288 x 2848) 2.7" vari-angle LCD Monitor Live View 720p HD Movie 4fps Continuous Shooting 19 Scene Modes

8 Calibration Both Nikons have been calibrated to a focus of 1.1m and their 24mm lenses have been fixed. This fixed focus will allow us to capture sharp and usable imagery between 0.9m 1.3m

9 CSI Camera and Lens IDs To maximise accuracy within the photogrammetric models, ensure imagery is taken using lens 1 on camera 1 and lens 2 on camera 2 - all have been labelled If the imagery becomes slightly out of focus, the focus ring may have moved let us know. Whilst there shouldn t be any need to recalibrate, the focus will need refixing to 1.1m

10 Image Coverage Using the project cameras, imagery taken at a distance of 1.10m from the object will cover an area approximately 1.0m x 0.7m If you take imagery at 0.9m and 1.3m the area covered will differ

11 Camera Settings - Image Quality and Size Image Quality: To ensure as much data as possible is collected, image quality should be set to NEF (RAW) + JPEG Image Size: set to Large 4288 x NEF/RAW automatically defaults to large The NEF/RAW format stores roughly 4 times as much info / data as the highest quality JPEG. The JPEG is captured for easy access to the image bank and doesn t require any specialist software or conversion. This may be revised (JPEG) capture is storage becomes an issue

12 Camera Settings Sensitivity Digital cameras have an ISO rating which can adjust the sensors sensitivity to light. Higher ISOs amplify the sensors sensitivity, so less light is needed. However noise is also increased, which is visible as graining and can impact on processing ISO Setting: Nikon is set to ISO 200, this should be more than adequate for the lighting conditions to be encountered. Try not to go above ISO 400

13 Camera Settings Shooting Mode The preferred shooting mode is Aperture Priority (A) this will allow greater control over the depth of field. Ideally the aperture should be set to f22, but a range from f11 f22 will be ok

14 Camera Settings Tripod and remote shutter Shooting at f11-f22 will slow the camera shutter speeds down and introduce the possibility of camera shake. To limit this, capture imagery using the tripod and the remote shutter The remote shutter is activated via the Release Mode menu on the D5000. Note: you need to be stood in front of the camera or at the side for it to fire

15 Camera Settings other info The 24mm lens provided with the camera only works in Manual Focus, it cannot be switched to Autofocus Turn off noise reduction, anti shake etc can impact on data stored and calibration

16 Imagery to be captured vertical The whole of the panels surface to be captured, not just the carved surface. This will be under constant review as the impact on storage and recording time may be significant Due to the shape of some panels, there ll be a mixture of different types of stereo photography, vertical, convergent and possibly oblique

17 Imagery to be captured vertical The English Heritage specifications for photogrammetric survey require that: Overlap between adjacent stereo images of at least 60% of the object area Overlap area of at least 60%

18 Imagery to be captured - oblique In addition to vertical coverage, oblique stereo-imagery should also be obtained of each carved stone This should ensure good stereo coverage is obtained of all identified rock art features, including any deeper, recessed hollows If time allows, a convergent, oblique stereo-pair should be taken from all four sides of the stone NB This type of coverage will also allow calibration of camera information lens distortion, focal length etc to be later derived. Often referred to as self-calibration this information is essential for accurate photogrammetric processing.

19 What needs to be in the image - scales Scales are required for accurate measurement to be applied to a model. Set the scales up at right angles to each other On large panels requiring multiple stereo pairs there s no need to have the scales on each pair. Include a scale where you can Ensure vegetation doesn t encroach over the sections on the scale The scales (or control points) should not be moved between the left and right image of the stereo pair. The slightest movement will cause the processing to fail

20 What needs to be in the image other Colour Chart to ensure a consistent colour is captured (or later matched), the IFRAO colour scale should be placed within each stereo-pair (off the area to be modelled) Additional targets or control points can be included in stereo pairs to aid the stitching process A north arrow isn t essential but may aid model orientation once processed

21 General Information A Minimum of 2 images referred to as a Stereo Pair are required to produce models An object / surface needs to be in view in both images for an accurate model to be recreated. If a surface can only be seen on one image the software will create a false surface between the areas it can see Using the Nikon D5000 we can expect accuracies of 1-2mm + depending on the number of images used and distance the imagery is shot at. 0.9m will produce greater accuracies than 1.3m, but more imagery is required Avoid shadows in motifs by using the parasol / umbrella. The software struggles to create surfaces in areas of dark or uniform colour without texture

22 Preparing the panel Ensure no vegetation encroaches into image, cut blades of grass back. Overhanging vegetation can cause spikes on the processed models Carefully brush off droppings / detritus in a downwind direction Soak up any water in the motifs with sponges and allow to dry Wet or muddy stones can cause issues during processing. You re best coming back later

23 Stereo Imagery Geometry Stereo imagery is comprised of a left and right image English Heritage suggest a Camera base to object distance ratio of 1:4 1:6 Others use 1:3 1:5 The distance moved between the left and right image is known as the camera base. The camera base is determined by the object distance

24 Stereo Imagery Geometry Move camera 0.18m 0.36m L to R 1.1m 1.1m Project cameras and 24mm lenses have been calibrated and fixed to a focus of 1.10m. Applying the guidance to 1.1m this equates to a base movement of 0.18m 0.36m

25 Stereo Imagery Geometry Move camera 0.15m 0.30m L to R A distance of 0.90m from the object 0.9m 0.9m There is a degree of flexibility within the 1.1m focus, allowing imagery to be captured between 0.9m 1.3m from the object

26 Stereo Imagery Geometry Move camera 0.21m 0.43m L to R A distance of 1.30m from the object 1.3m 1.3m There is a degree of flexibility within the 1.1m focus, allowing imagery to be captured between 0.9m 1.3m from the object

27 Types of stereo imagery Geometry The English Heritage specifications for photogrammetric survey require that: X A variation in the scale between adjacent stereo-images of no more than 5%. At the focus distance the project is working at (around 1.1m) the English Heritage specification equates to a difference 5-7cm between the left and right image

28 Stereo Imagery Geometry The English Heritage specifications for photogrammetric survey require that: Alignment of each image plane with the principal plane of the object area to within +3 degrees of parallelism Object plane Image plane

29 Stereo Imagery Geometry Plan View X Keep as parallel to the stone as possible when taking the imagery

30 Image Geometry Oblique Tilt The English Heritage specifications for photogrammetric survey require that: Alignment of each image plane with the principal plane of the object area to within +3 degrees of parallelism In other words try to maintain the same camera tilt for each oblique stereo-pair

31 Image Geometry Oblique Tilt The English Heritage specifications for photogrammetric survey require that: Minimised vertical tilt of the camera in either an upwards or downwards direction, up to a maximum of +15 degrees In other words try to limit the downwards (or upwards) tilt of the camera for each oblique pair

32 Types of stereo imagery vertical Remember the camera and 24mm lens calibrated to a focus distance of 1.1m Maintain the same distance between the camera and object /stone when moving left to right on the stereo pair

33 Types of stereo imagery divergent Images taken with camera angled away from object X A model can be produced with this imagery, but the surface accuracy may be reduced

34 Types of stereo imagery convergent Images taken with camera angled in towards the object But remember base ratio is still important X In reality it will be impossible to avoid taking convergent imagery on some panels due to the stones shape A model can be produced with this imagery, however the accuracy may be reduced

35 Multiple stereo pairs large panel Photo 1 Photo 2 Photo 3 Photo 4 Photo 5 Photo 6 Photo 7 Rock art panel Ground

36 2.5m 0.7m Plan view of a flat ground level stone

37 Leave a gap on the left side of the shot to allow for the L to R base movement L Targets placed near the right edge of the SP left image can guide overlap between the stereo pairs Base

38 L R m

39 Area covered by the first stereo pair

40 On the left image of the second SP leave enough overlap to allow for movement to the right L The scale bar (s) and IFRAO scale can be moved to subsequent stereo pairs Take the next SP by moving the camera / tripod right along the base, remembering to keep parallel to the panel. Aim for a minimum of 20% overlap to the area covered by the previous stereo pair

41 L Place another set of targets near the right edge of the left image

42 The 2 SPs will be joined together by this area of overlap L R m

43 Area covered by 2 nd SP L R

44 Area covered by 2 nd SP L L Repeat same procedure for the 3 rd stereo pair Remember this movement is determined by the required overlap not the base / height ratio

45 L L m R

46 L

47 Multiple stereo pairs strips Targets to aid stitching during processing Large panel (plan view) Image strip Image strip tape measure used as a guide Image strip

48 Multiple stereo pairs strips - Hangingstones

49 Hangingstones - model 5-10mm accurate surface

50 Image Capture Summary Maintain the distance between the camera and object on all stereo pairs Remember the overlap Avoid excessive tilt between stereo pairs Carefully brush off droppings / detritus in a downwind direction Soak up any water in the motifs with sponges and allow to dry Wet or muddy stones can cause issues during processing. You re best coming back later

51 PhotogrammetryRecording Form RF4 RF4 is used to record the photogrammetry image filenames for each panel whilst out in the field Provides important positional information about the imagery useful when processing A reference to sort images easily when uploading to the computer A useful document if image files are misfiled when copied to the hard drive / computer Forms part of the future record The form isn t perfect for every panel, suggestions are welcome

52 Photogrammetry Recording Form RF4 This number is generated by the ERA database once the record has been input This section records the camera / lens used. Enter 1 or 2 Note: you should have camera 1 and lens 1, or camera 2 and lens 2, not camera 1 and lens 2. Enter approx distance (m) the imagery was taken at Note: if issues are encountered during processing, we may be able to pin down the problem to a distance (too near or too far)

53 Photogrammetry Recording Form RF4 Roughly sketch an outline of the panel and a few of the motifs Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end N DSC1001 DSC1002 DSC1003 DSC1004 DSC1005 DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012

54 Photogrammetry Recording Form RF4 For small panels fill in the top part of the table Enter the image filenames for the stereo pair in the table ensuring cross reference is easy Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end N DSC1001 DSC1002 DSC1003 DSC1004 DSC1005 DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012 Mark the location and direction the imagery was taken from

55 Photogrammetry Recording Form RF4 Note the two SPs taken here were vertical, hence the column Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end N DSC1001 DSC1002 DSC1003 DSC1004 DSC1005 DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012 Continue to mark the location and direction the imagery was taken from

56 Photogrammetry Recording Form RF4 Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end N DSC1001 DSC1002 DSC1003 DSC1004 DSC1005 DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012 Continue to mark the location and direction the imagery was taken from

57 Photogrammetry Recording Form RF4 Same procedure is followed for large panels DSC DSC DSC Row Row Row

58 Photogrammetry Recording Form RF4 Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end Please feel free to rename columns and rows, as long as we re able to easily cross reference filenames to position N DSC1001 DSC1002 DSC1003 DSC1004 DSC1005 DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012 Due to the complexity and number of images being captured during this project, the table layout will not work for every panel

59 Photogrammetry Recording Form RF4 Context view looking W Context view looking E Context view looking S Context view looking at Haystack Close up of motifs at S end View of peck marks S end DSC1001 DSC1002 DSC1003 DSC1004 Complete as appropriate. If No please indicate reasons in Comment DSC1005 N DSC1006 DSC1007 DSC1008 DSC1009 DSC1010 DSC1011 DSC1012 Enter filenames of processed models & formats used Enter the surface measurements applied to the photogrammetric model(s)

60 Now your turn