Introduction

People are usually far better at remembering interactive rather than static material. Interaction with knowledge develops deep learning rather than textbook memorization, thus students can benefit greatly from the introduction of interactive technology. In the past few years many Universities and Colleges have produced their own World Wide Web pages with links to teaching material, papers, exercises etc., so that the rest of the world can benefit from their work. This idea has had a great response, resulting in huge amounts of information becoming available to any user with Internet access and a browser. The typical educational WWW page is a hyper-text document written in HTML, sometimes containing multimedia objects such as embedded images, sounds, demonstrations, etc. However, what is needed is a real hands on application, for which JAVA is ideal. What we ought to have are ``interactive textbooks''.

The project described here explores the potential for JAVA to implement and link a variety of image processing operators within a HTML hypertext document, for the purpose of creating ``interactive textbooks''. The project explored functional (i.e. ``Does JAVA have enough programming facilities?''), practical (i.e. ``Is JAVA fast enough and does it have enough computational resources?'') and educational (i.e. ``Is the combination usable, attention-getting and informative?'') issues. The final product can be viewed using a JAVA-enabled web-browser and the operators communicate by passing on output data to each other.

The idea of an interactive teaching package is common, and has even been applied to image processing [15], however, projects are usually limited to use on the platforms for which they were designed, which, as in the case of [15], may become obsolete very rapidly. The advantage of using JAVA is that it is platform independent and seems likely to be publically available for a long time.

To investigate the suitability of JAVA for use in interactive image processing textbooks, representatives of the main classes of computations typical to image processing were chosen for implementation. They include:

Point Operators

are applied to individual pixels and are, therefore, position independent operators. Two popular representatives of this category are Thresholding and Gamma correction.

Image Arithmetic

combines two or more images to produce a single one as an output. Image subtraction, and logical AND/OR are common examples.

Geometric Operations

, e.g. Rotation, Translation and Scaling - affecting the position but not the content of the image data.

Morphology

take as input a binary image and a structuring element, and output a function of the two - linked by the relative spatial distribution of the pixels, rather than their values.

Digital Filters

are often used for smoothing or enhancing features in images and are based on a two-dimensional convolution operation (which expresses a linear filtering process applied to an image). The convolution of two functions f and g is defined by

where represents the image and the kernel (see [13]).

Our main conclusions are that, in general, JAVA is suitable, integration with HTML is straightforward, the execution time of interpreted JAVA is sub-optimal but compiled JAVA performs satisfactorily. One has to take care not to misuse the internal thread resources of JAVA as the browser implementing the JAVA Virtual Machine may become overloaded.