Geospatial services

Orthorectification for Precision Mapping



Planimetrically accurate images are a necessary pre requisite for any mapping projects.  Accurate maps are a basic necessity of the developmental projects like, infrastructure development, mining, master planning, precision agriculture etc. Orthorectification is a process by which a raw image is transformed into a planimetrically accurate image.  Often georeferencing is used in synonym with the orthorectification.  But in true sense, through georeferencing only x, y corrections and transformation is dealt with while ortho rectification will do a correction on relief displacement caused due to elevations of the objects within the map.  So orthorectification is considered as most accurate process by which an image is brought to the real world coordinates by considering correction in all three directions of x, y and z.

Orthorectification is the process of eliminating the relief displacement due to sensor mal functioning, height of the terrain, earth rotation, panoramic distortions etc.  The outcome of the orthorectification is a planimetrically true image.  It also brings uniformity in scale throughout the image.    SBL has the capability to do the orthorectification to a wide variety of images such as satellite images, aerial photographs and UAV images.

The main input for the orthorectification process is as follows

  • Near Nadir image
  • Rational Polynomial Coefficient (RPC) file or camera calibration certificates
  • Digital Elevation Model (DEM)
  • GCPs

Orthorectification can be done in a variety of software.  SBL uses ortho master module of Inpho for the aerial photographs and satellite images and PIX4D for UAV images.  The process involves importing of the image, DEM, external and internal parameterization files in to the model.  Parametric approach can be followed in the model where internal and external correction files in the form of GCP files and RPC files are involved.  At this step desired projection can also applied in the model.  Back ward projection can be applied in the process.  The object space X, Y coordinates related to every pixel of the final ortho image is determined. The height Z at a specific X, Y point will be calculated from the DEM or the 3D model and then the X, Y, Z object space coordinates are projected in the original image in order to acquire the gray level value for the ortho image pixel. Interpolation or re-sampling process in the original image is also essential because of the fact that the projected coordinates will not fit accurately at the original image pixel centers.  Output cell size and extent will be defined at this step.  Output file type can also be specified at this step.  Error calculation is generally derived in the form of RMSE report.

Quality assessment of the orthorectified images is one of the activities which SBL perform to ensure highest standard of the output images.  Pan sharpening, mosaicking, color balancing, and tiling are the associated activities once the generalized orthorectified output images were generated.    Quality assessment will be majorly done through visual inspection of the images.  Major checks on the orthorectified images include spatial resolution, edge matching, double feature checking, tonal balance, and pixel dropouts.

Satellite Image Interpretation for Land Cover Classification

SBL’s Geospatial wing recently won a project from Canada for the delineation of Land cover in Northwest Territory. This is again a reaffirmation of the varied skills of the Geospatial team wherein our pool of experts were able to provide a better solution to the forest department based on our quality of interpretation of the sample data and competitive pricing.

Land cover basically refers to the interpretation of land features that are visible on the surface of the earth. It does not cover the land usage which comes under Land use. 

Land cover classification includes land features like trees, water bodies, bare earth and rock outcrops. In the present study our task was to interpret Land cover in forested regions of Northwest Territories, Canada. The input image provided by the client was mono satellite image and each image covered an area of 4

SBL’s experienced interpreters delineated the treed and non treed areas and created polygons out of it. The client had specified a minimum area for forest polygons and the area specified was 4 hectare and for non forested area polygon it was 0.5 hectare. The classification was done based on the tone and texture of the features(Fig.1).

Forest delineation based on tone and texture

The tree heights, texture and tree density were taken into account for the delineation. Trees situated in the high land regions will have a different tone, texture, height and density with respect to the trees in the high land and mid land regions(Fig.2, Fig.3).

Landcover classification in highland region

Landcover classification based on density of vegetation

Various tree species occurring in the forest areas display unique textural and tonal characteristics which were used as guide for the interpretation.

The water bodies consisted of rivers, major streams and also the closed water bodies like lakes which fall within the area stipulation of 0.5 hectares were also captured (Fig.4).

Flood plain delineation

Rivers were captured at the edge of water and flood plains were delineated wherever the width is greater than 20m. The digitization was to be done in a scale level of 1:2000 and 1:2500 and the final output was viewed in a scale of 1:10000 in order to find out any dangle polygons, large polygon and also to merge adjacent polygons if necessary. The entire digitization was carried out in ArcGIS environment and the final output was delivered as shape files to the client.