Perspective Data/Information

Perspective Data / Information is information obtained through a perspective process or derived from a perspective view, image, representation, model, measurement or calculation in relation to spatial reality. It concerns what perspective enables us to know, identify, compare, measure or infer about objects, scenes, images and spatial relationships.

As a Perspective Product, Data / Information extends the idea of perspective beyond the production of visible images. A perspective process can also produce knowledge about the spatial reality represented by those images.

Data / Information as a Perspective Product

The Product level asks what a perspective process produces. In addition to a visible View, Image, Representation or Model, the process may produce or enable the extraction of spatial information.

This information may be directly visible, measured, calculated, compared or inferred from the relationship between a Perspective Image and the spatial reality to which it refers.

Three Principal Kinds of Perspective Data

The PRC framework distinguishes three broad kinds of Perspective Data:

Subsumptive Data — information concerning whole and part relationships.
Ordinal Data — information concerning figure, shape, size, scale and position.
Determinative Data — information concerning state or activity.

These provide a useful framework for describing the information that may be obtained from perspective images, views, measurements, calculations and models.

Subsumptive Data — Whole and Part

Subsumptive Data concerns the relationship between wholes and their constituent parts.

A Perspective Image may allow an observer or technical system to determine whether a visible element belongs to a larger object, how components are grouped, which parts overlap or contain others, and how local structures relate to a larger spatial arrangement.

This kind of information is important in visual recognition, image interpretation, scientific observation, technical inspection and computer vision.

Ordinal Data — Figure, Size, Scale and Position

Ordinal Data concerns spatial properties such as figure, shape, size, scale and position.

A perspective view or image can provide information about relative position, apparent dimensions, orientation and spatial arrangement. With sufficient calibration or additional geometrical information, such data may also support quantitative measurement and reconstruction.

The accuracy and meaning of the resulting information depend upon the perspective system and the conditions under which the image or measurement was produced.

Determinative Data — State and Activity

Determinative Data concerns the state or activity of an object, scene or process.

A sequence of perspective images may reveal movement, change of position, deformation, growth, rotation or other activity. Even a single image may sometimes provide visible evidence concerning state, condition or action.

Motion-picture imaging, surveillance, scientific imaging, remote sensing and computer vision frequently depend upon this class of perspective information.

What Information Can a Perspective Image Provide?

Depending upon the perspective system and available evidence, a perspective view or image may provide information concerning:

location;
size and shape;
scale and proportion;
position and arrangement;
orientation and direction;
depth and distance;
focus and visibility;
surface and texture;
colour and brightness;
state, movement and activity.

Not every image provides all of these reliably. What can be recovered depends upon viewpoint, projection, resolution, optical conditions, occlusion, calibration and the information available from other views or measurements.

Perspective Phenomena as Information

Many Perspective Phenomena carry information about spatial relationships.

Diminution can contribute information about relative distance; foreshortening can reveal orientation; overlap or occlusion can establish ordering in depth; convergence can indicate directional relationships; and changes in colour, clarity or contrast can contribute to judgements of depth and atmospheric distance.

A visible Perspective Phenomenon is therefore not merely an appearance. It may also function as evidence from which spatial information can be extracted.

Data Depends on Perspective Conditions

Perspective information is always obtained under particular conditions. Viewpoint, viewing direction, projection method, image scale, field of view, focus, resolution, occlusion and other factors affect what information is visible and how reliably it can be interpreted.

The same spatial object can therefore produce different Perspective Data when observed from different positions, through different imaging systems or at different scales.

The information is not necessarily false; it is conditional upon the perspective relationship through which it was obtained.

From Image to Measurement, Calculation and Information

A perspective process can produce a sequence of related Products.

A Perspective Image may first be formed or captured. Features within that image can then be measured. Those measurements may enter a calculation, and the resulting values may finally become information about spatial reality.

These stages are analytically distinguishable even when they occur automatically within a single technical or computational system.

Direct and Inferred Information

Some perspective information is comparatively direct. A visible object may clearly lie in front of another, or a calibrated scale may provide a measurable dimension.

Other information must be inferred. Depth, physical size, distance, orientation or three-dimensional form may require assumptions, geometrical reconstruction, comparison with known objects, multiple viewpoints or additional sensor data.

It is therefore important to distinguish what is directly present in an image from what has been reconstructed or inferred from it.

Multiple Views and Additional Information

A single Perspective View reveals only part of the information potentially available about a three-dimensional object or scene.

Additional viewpoints can reveal previously hidden surfaces, resolve ambiguous depth relationships and provide further positional or geometrical information. Stereoscopic, multi-view, moving-camera, photogrammetric and computer-vision systems exploit this principle.

A Perspective Model can similarly provide many different views of the same underlying spatial structure, increasing the amount of accessible information.

Examples of Perspective Data / Information

determining which object lies in front of another;
identifying whole and part relationships;
estimating relative depth or distance;
determining object position or orientation;
measuring size or scale from calibrated imagery;
identifying movement or change through successive images;
recovering spatial coordinates from multiple views;
classifying visible objects or surfaces;
detecting changes in shape, position or state;
deriving information about spatial reality from camera, satellite or scientific imagery.