School of Computer Science Page 1 of 2

COMP5405/4405 Digital Media Computing 2024S1

Homework

Homework is an individual assessment and is worth 15% of the total assessment of this unit

of study. It is due in Week 08.

You need to

1) make an online submission of your source codes and a brief report of your

implementation.

Due: Mon, 15/04/2024 (Week 08)

2) demonstrate your work to your tutor during lab time. Marks may be deducted if you

are not able to explain your work clearly.

Task –Channel alignment of color images

As we all know, a color image can contain 3 channels, red, green, and blue. This feature is

utilized by some photographers in the 1900s to generate color photos from monochrome

cameras.

The photographer will firstly apply a red filter before the camera, only allowing red light to

expose the film, recording the brightness of only the red light (Fig. 0). Then he/she will do

the same procedure with a blue filter and a green filter. Therefore, 3 photos containing the

brightness information of the 3 colors. The photographer will then he/she can present the

photo with color by a specialized project. However, due to some distortion of movement of

the cameras when taking different shots, 3 channel images can be displaced, and it leads

to inferior quality of combined images (with serious ghosting) when simply stacked.

Fig. 0 Camera capturing only the red lights

In this homework, you will be given three individual channels of brightness information of

several images, and you need to build a program to combine three channels with alignment

adjustment so that it resembles the original picture.

An example is given where Fig. 1 is the original photo picture, Fig. 3 is the three channels,

and Fig. 2 is an undesired simple stacking result.

Fig. 1 A desired color photo Fig.2 Simple Stacking Results(undesired)

A Red Filter

COMP5405 Digital Media Computing HOMEWORK

School of Computer Science Page 2 of 2

Fig. 3 Brightness Images of 3 channels

Marking Scheme (15 marks)

1. [12] marks] Algorithm implementation

Your submission will be given six marks if your program is able to produce a result

reasonably similar to Fig. 1(b).

An implementation of an algorithm to produce a result comparable to Fig. 1(b) will be

given up to 9 marks.

An implementation of an advanced and recent algorithm to produce a result comparable

to Fig. 1(c) will be given up to 12 marks.

Some possible algorithms are listed in References.

2. [3 marks] Report

A brief report up to 5 pages to describe your method and results. It could contain key

steps or details of the algorithm implemented, experimental settings, and experimental

results, and several key references. A clear report will allow readers to reproduce your