Classification Function

---

Classification Function concerns the use of perspective information to identify, compare, distinguish, group, match and systemise objects, scenes, images, spatial properties and Perspective Phenomena.

It is particularly important where a perspective process is used not merely to see or measure something, but to determine what it is, which properties it possesses, how it relates to other objects or images, and whether it corresponds to an established class, standard or reference.

---

Classification as a Perspective Function

A perspective system performs a Classification Function when information derived from a scene, view, image, measurement or model is used to distinguish one spatial form, property or condition from another.

The operation may be visual, graphical, optical, mathematical, instrument-based or computational.

Classification can therefore range from ordinary human recognition of objects within a scene to automated identification and sorting performed by computer vision or machine-vision systems.

---

Principal Classification Operations

Within the PRC Functions of Perspective map, Classification includes five principal operations:

• Survey — inspect the relevant properties of a scene, object or image;
• Compare / Identify — determine similarities, differences or identity;
• Match — relate a scene, object or image to another object, image, standard or known condition;
• Cross-match — compare information derived from multiple images, views, systems or sources; and
• Prescribe / Certify — establish whether specified properties or conditions satisfy a defined requirement or standard.

---

Surveying Before Classification

Classification normally begins with some form of observation or survey.

A scene, object or image must first be examined so that relevant features can be isolated. These may include shape, size, location, orientation, colour, clarity, depth, surface structure, spatial arrangement or Perspective Phenomena.

Surveying therefore provides the descriptive information upon which later comparison and identification may depend.

---

Compare and Identify

Comparison determines whether spatial or visual properties are similar, different or related.

Identification goes further by associating those properties with a recognised object, form, type, condition or reference.

A perspective image may therefore be used to identify objects from outline, apparent shape, scale, colour, texture, spatial position or combinations of several visible properties.

---

Matching

Matching relates the properties of one image, view or spatial reality to another known or measured reference.

This may involve comparison with:

• another perspective image;
• a previously recorded view;
• a measured spatial object;
• a graphical or geometrical model;
• a known scale or standard;
• a database of recognised forms; or
• a prescribed set of spatial properties.

Matching therefore provides an important connection between perspective information and knowledge about the spatial reality being examined.

---

Cross-Matching Multiple Views

Cross-matching compares corresponding information across two or more images, views or data sources.

This is especially important in systems where no single perspective view contains sufficient information to establish the required spatial relationship.

Multiple photographs, stereoscopic image pairs, sequential camera views, remote-sensing images or computer-generated views may be compared to identify corresponding objects, features or spatial relationships.

---

Segmenting and Partitioning Spatial Reality

Classification often requires a scene or image to be divided into meaningful parts.

A system may segment a visual field, pinpoint individual objects, distinguish figure from background or partition spatial reality according to particular properties.

This allows complex visual information to be converted into identifiable objects, regions, groups or relationships that can then be compared and classified.

---

Prescribing and Certifying

Classification may also operate in the opposite direction: instead of merely identifying what exists, a perspective system may establish what properties should exist.

Prescribing establishes required visual, spatial or geometrical conditions. Certifying determines whether the observed or measured result satisfies those conditions.

This is relevant wherever perspective information is used for inspection, calibration, verification, quality control, surveying, mapping or technical assessment.

---

Classification and Measurement

The Classification Function is closely related to the Measurement Function, but the two are not identical.

Measurement determines or compares quantities and spatial relationships. Classification interprets such information by identifying, grouping or matching the resulting properties.

For example, a system may first measure the apparent dimensions of an object and then classify the object by comparing those dimensions with known examples.

---

Classification and the Match Goal

Classification belongs principally beneath the broad Match family of perspective functions.

The broader Match goal concerns relating images and views to known spatial forms, standards, impressions or measuring scales. Classification provides many of the operations through which such matching becomes possible.

Measurement and Classification therefore form complementary parts of the Match family: one establishes properties and relationships; the other identifies and organises them.

---

Human and Machine Classification

Classification may be performed by a human observer or by a technical system.

Human visual classification may involve recognising an object, distinguishing one spatial form from another or judging whether an appearance corresponds to a known condition.

Technical and computational systems may perform comparable operations through image processing, pattern recognition, computer vision, remote sensing, automated inspection or machine learning.

In each case, perspective information provides part of the evidence from which the classification is made.

---

Classification Function and Perspective Classification

The Classification Function should not be confused with the analytical classification of perspective itself into Categories, Classes, Types, Forms, Phenomena, Functions and Products.

The analytical system classifies perspective concepts. The Classification Function describes an operation through which a perspective process classifies objects, scenes, images, properties or spatial relationships.

The same word is therefore being used at two different conceptual levels, and the distinction is important.

---

Classification Within the Analytical System

Classification belongs to the Function level of the seven-part analytical system:

• Category — principal source or mode;
• Class — fundamental direction;
• Type — method, system or process;
• Form — resulting appearance;
• Phenomenon — visible or detectable effect;
• Function — what the process does; and
• Product — what the process produces.

Classification is therefore an operation that may occur within many different Perspective Categories, Classes and Types rather than constituting an additional Category or Class of perspective.

---

Related Perspective Topics

Functions of Perspective →
Matching — Goal of Perspective →
Measurement Function →
Observation Function →
Capture / Project Function →
Perspective Process →
Perspective Phenomena →
Perspective Category →
Perspective Class →
Perspective Type →