Photogrammetry Workflow Tutorial Step by Step for Student and Professionals

Learn complete photogrammetry workflow steps which involved in any digital photogrammetry application, specially for students and professionals. The process includes Scanning, Orientation, Epipolar Resampling, Digital Terrain Modelling (DTM), Digital Planimetric Mapping and Ortho Photo Generation.
photogrammetry work flow

Scanning :
Digital Photogrammetry begins with image scanning, which converts a photographic image into a digital image file. A scanner is a precisely guided CCD camera (a camera with a charge-couple-device installed in the film or image plane), which traverses over the entire 23 x 23 cm image format. A scanner can also be precisely controlled moving stage that presents the film image to a fixed CCD camera.

The lens in a scanner’s CCD-camera can enlarge or reduce the image in view. However, the optical resolution of a scanner (commonly referred to as the optical scanner resolution) is ultimately limited by the element size in the CCD chip installed in the camera image plane. A digital image is composed of pixels of fixed size, shape and spacing, but which have variable brightness values. If a scanner has an optical resolution of 10 µm (or 2,540 dpi) and outputs 8-bit files, then pixel brightness values from 0-255 are produced in a rectangular array, spaced 10 µm apart.

Coarser scanner resolution can be achieved (regardless of the scanner optical resolution) by sampling the film image during scanning, or by resampling the optically resolved scan pixels (Ps) into other sizes using averaging or other image convolution techniques. Coarser resolutions can be obtained in multiples of Ps if the image is sampled during scanning, or in any pixel size is obtained resamplling.
During scanning processes, the gray value or the color value of the original document is measured by a photosensitive element, either by a photo multiplier or by a photo multiplier or by semiconductor image sensors. When treating black and white photographs one generally discreticizes the hue into 256 grey values coded on 8  bits.

A colour film is generally digitized in 3 bands (red, green, blue) using color filter. Most of the films are composed of three layers dyed according to the principle colors of subtractive color mixture (yellow, magenta, cyan). The information content of the sand which film is measured by rather narrow filters in the colors red, green and blue for color separation, again using 8 bits for each band.

Interior Orientation: It is the process of measuring the fidicial marks on the digital imagery. This measurement process will relate the pixel coordinates of the digital image to the photo coordinate system. Once the fidicials on the first photo have been measured, all the fidicials on subsequent photos will be automatically measured using image correlation techniques.
Relative Orientation: It is the process of measuring (removing x and y parallax) at least 6 points, in the overlap region on both the left and right photo of a stereo pair. This measurement process will relate the two photos to each other in space and transform the photo coordinates of the measured points into model coordinates.

The point measurement process is useful to measure in both mono and stereo modes at the same time. All pass points (points tying stereo models together), tie points (points tying strips together), and control points can also be measure during the relative orientation step. If aerial triangulation adjustments are required, all measurements from the relative orientation step can be output to the adjustment package. After adjustments have been made, the results are read back into the environment where the exterior orientation parameters will be used to epipolar resample each stereo model.

Absolute Orientation: If aerial triangulation is not required (because there are at least 3 control points already identifiable in each model), then the process of absolute orientation will be the next step. If all pass point s have to be checked off and the software will calculate the elements of exterior orientation for each photo relative to ground control.

If these points have not yet been measured (x parallax removed), then they can be measured and or transferred in the absolute orientation environment in the same way as they would have been measured in relative orientation. This means measurement either in mono or stereo or both. This can be accomplished either manually or automatically with the help of image correlation and common point locking.

Benefits of orientation:
Point transfer and measurement directly on digital imagery
Capability to tie large blocks and strips of photos
Common point lock to aid point transfer
Automatic measurements using image corelation techniques
Manual measurements can be made in mono, stereo or both
Single photo resection and orientation available
Bundle adjustment
All orientations are a “one-time” procedureable to be recalled at anytime with a push button

Epipolar Resampling:
After the elements of exterior orientation have been calculated for each photo, either from the aerial triangulation adjustment or directly form the aerial triangulation adjustment or directly form absolute orientation, the photos will then be resampled into epipolar format.
Resampling to the epipolar format involves a rearrangement of the pixels in each digital photo of a stereo model in such way as to remove all the y parallax (i.e. along epipolar lines). Epipolar resampling also removes the effects of tilt, tip, and awing from the aircraft and incorporates any corrections the operator may have applied such as earth curvature, atmospheric refraction and lens distortion. This is preferred viewing mode in digital photogrammetry, significantly reducing eye strain and therefore operator fatigue. The resulting images will then be ready for stereo feature and
DTM collection.

Digital Terrain Modelling (DTM)
Digital terrain models can be collected interactively with softcopy solution. The on-line triangle and contour generation solution allows the operator to view and manipulate, in real-time, the contours or triangles oner the collected area while staying within the model.

DTM Point Density should be based on the nature of terrain surface and the degree of topographic resolution required. The automatic DTM generation tools will collect a DTM at user-specified grid spacing in less than 20 minutes; this saving in time (as compared to collecting the same area analytically) is one of the significant reasons why most photogrammety and mapping organizations are going for digital photogrammetry today.

Operator training is easier in digital photogrammetry environment than on any analytical environment. It is very easy to setup training files that are small yet very specific (e. g. urban areas, mountainous areas, dense vegetation areas, etc.) and have a supervisor observe at anytime that work being done on the stereo monitor without interrupting the operator.

DTM Benefits
Improved operator comfort and greater productivity
Stereo superimposition of 3D vectors
Group stereo viewing for training and supervision
Interactive DTM collection
On-line triangle and contour generation
Image correlation for assisted measurement

Digital Planimetric Mapping
Planimetric mapping (or feature compilation) is the data capture task by interpreting and highlighting image objects in an oriented stereo model. The result is a data file, which contains the digital coordinates of the various interpreted objects. The file is considered a Positional File where the locations (X,Y,Z) of the mapped objects are important rather than their appearance in a map/image. Objects or features are symbolized by simple point, line or aerial vectors, often with alpha numeric code (feature code) attached to each vector object for identification.

Ortho Photo Generation
The output from the DTM generation process of the above step is used as input into a terrain analysis module. The DTM will be merged with any other terrain information like breadlines, spot height etc. to generate either a “TTN” or “GRD” file. A TTN is a topologically triangulated network, often referred as a TIN or triangle file. A GRD file is a raster grid file that stores X, Y and Z values for regularly spaced posts. The softcopy solution can use either of these popular formats for the ortho resampling process. The terrain analyst module can be used to analyse plan or perspective views of topographic features, triangles, surface mesh, contours, and color-coded elevations. The terrain analyst module is having the capability to represent 3D-terrain information in a large number of projection and coordinate systems.

After the DTM file has been generated in Terrain analyst module, all the elements are in place (original digital image, elements of exterior orientations, DTM) for creating a digital orthophoto. The solution is capable of orthorectifying digital imagery. Any ground features collected in stereo can be displaced in their true orthographic positions directly on top of the newly generated orthophoto. Additional digitizing in 2D can also be performed directly on top of the digital ortho photo.

Once the orthophoto has been generated, it can be viewed in any image processing softwere. The image processing software can be used not only for display but also to do enhancements, including tone matching and seamless mosaicking of overlapping orthoimages. It achieves the seamless result during mosaicking by applying a feathering algorithm on either side of the seam line in the overlap area between two digital images. The seam can either is automatically generated can use an existing design file element (e.g. a road, rive boundary etc.). The feathering width is adjustable and does an excellent job of blending one photo into the next without any visual trace of the join.