Current Documentation
View previous documentation
View legacy documentation
All documentation versions

Deprecated in Current Release

Requirements Software operation Defining Calibration Tasks Definitions and Theory

Supported Systems

Supported Targets

Required Evidence

Test Setup

Calibration Procedure

User Interface

Use in Imatest IT

Module settings

Module outputs

Defining a Device

Defining Distortion

Defining the System of Devices

Defining the Target

Defining a Test Capture

Defining a Test Image

Homogenous Coordinates

Projective Camera Model

Multi-Camera Systems

Distortion Models

Coordinate Systems

Rotations and Translations


A Device is synonymous with an individual camera- a lens and sensor together producing images. Typically the values supplied in this definition come from nominal device specs.

Geometric Calibrator requires some knowledge of the geometry of the sensor, which is necessary to provide a precise mapping from pixel array locations to precise physical locations on the sensor plane.

An initial estimate of a functional description of the lens is also required. These values, Effective Focal Length (EFL) and distortion model description, are used as a starting point for calibration optimization and estimated actual values for the device (as opposed to from spec) are returned by the calibration routine.

The relevant required parameters are illustrated in Figure 1, below. 

Figure 1, Illustration of required Device definition parameters


Sensor Size

Supported values: 2-array

Size of images produced by the device, [nRows, nColumns]. 

Note: Differently scaled or cropped images provided from a sensor require a different Device definition, in order to maintain a proper definition of spatial coordinates on the sensor plane. It is therefore essential that you provide here the size of the actual image format in which you will be providing images to the Geometric Calibrator later. 


Pixel Pitch

Supported values: scalar value, with accompanying pixel pitch units property

The regular spacing between the pixel locations on the sensor, typically provided in microns. 

Note: If a sensor has non-square pixels, supply one dimension’s pitch here and enable alpha estimation in the calibration- this will the estimation of the relative sampling rates (i.e., pitch) in each direction.


Lens Type

Supported values: RECTILINEAR or FISHEYE

This flag is a coarse classification which simply indicates to the software that it may need to work with images from this device specially (e.g. for target detection in images from devices with fisheye lenses). Default is RECTILINEAR.



Supported values: scalar value, with accompanying efl units property 

The Effective Focal Length of the thin-lens approximation of the device optics, typically provided in mm.

This value is provided as an initial estimate and should be known from the lens component spec sheet. The actual EFL of the particular unit under test will be estimated. 


Distortion Model

Supported values: a Distortion Model definition

An initial estimate of the distortion model describing the device’s optics and how they break with the rectilinear pinhole camera model. This is also used as a declaration of the distortion model to fit during calibration- the model type and number of parameters of the final calibration will match this initial input form.


Image Circle Radius

Supported values: non-negative value

Maximum distance from image array center to use for detecting target features. Useful for some fisheye lenses which produce an image only on a circular sub-region of the whole sensor. If not supplied, the whole area is used in any images from this device.