UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Lazy Image Processing: An Investigation into Applications of Lazy Functional Languages to Image Processing

Kozato, Yasuo; (1994) Lazy Image Processing: An Investigation into Applications of Lazy Functional Languages to Image Processing. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[img]
Preview
Text
Lazy_image_processing_An_inve.pdf

Download (21MB) | Preview

Abstract

The suitability of lazy functional languages for image processing applications is investigated by writing several image processing algorithms. The evaluation is done from an application programmer's point of view and the criteria include ease of writing and reading, and efficiency. Lazy functional languages are claimed to have the advantages that they are easy to write and read, as well as efficient. This is partly because these languages have mechanisms to improve modularity, such as higher-order functions. Also, they have the feature that no subexpression is evaluated until its value is required. Hence, unnecessary operations are automatically eliminated, and therefore programs can be executed efficiently. In image processing the amount of data handled is generally so large that much programming effort is typically spent in tasks such as managing memory and routine sequencing operations in order to improve efficiency. Therefore, lazy functional languages should be a good tool to write image processing applications. However, little practical or experimental evidence on this subject has been reported, since image processing has mostly been written in imperative languages. The discussion starts from the implementation of simple algorithms such as pointwise and local operations. It is shown that a large number of algorithms can be composed from a small number of higher-order functions. Then geometric transformations are implemented, for which lazy functional languages are considered to be particularly suitable. As for representations of images, lists and hierarchical data structures including binary trees and quadtrees are implemented. Through the discussion, it is demonstrated that the laziness of the languages improves modularity and efficiency. In particular, no pixel calculation is involved unless the user explicitly requests pixels, and consecutive transformations are straightforward and involve no quantisation errors. The other items discussed include: a method to combine pixel images and images expressed as continuous functions. Some benchmarks are also presented.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: Lazy Image Processing: An Investigation into Applications of Lazy Functional Languages to Image Processing
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Thesis digitised by UCL
URI: https://discovery.ucl.ac.uk/id/eprint/10102891
Downloads since deposit
0Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item