Compression

Lossless compression reduces the file size by storing the data more efficiently so that no data is lost, and the exact original file can be reconstructed from the compressed version.

1. A smaller file is transmitted faster and uses less bandwidth.
2. A smaller file uses less storage space on the server and the receiving device.

Lossless compression reduces file size without losing any data; the original file can be reconstructed exactly. It is used where the data must not change — for example text, spreadsheets and program files. Example file types: PNG, FLAC, ZIP.

Lossy compression reduces file size by permanently removing some data (usually detail the human eye or ear is less sensitive to). The original cannot be perfectly reconstructed, but file sizes are much smaller. It is used for photos, music and video where some quality loss is acceptable. Example file types: JPEG, MP3.

(a) Identify the runs: R occurs 4 times, G occurs 6 times, B occurs 4 times. As count–value pairs: 4R 6G 4B. (14 pixels stored as 3 pairs.)

(b) RLE is particularly effective for images with large areas of a single, repeated colour — for example simple icons, logos, line drawings or cartoons.

(a) X-ray images — lossless. Diagnostic detail must be preserved exactly; removing data could hide a feature that affects diagnosis, so no loss can be tolerated even though files are larger.

(b) Holiday photos on social media — lossy. A small loss of fine detail is not noticeable on a phone screen, and the priority is small files that upload quickly and use little storage/bandwidth, so lossy (e.g. JPEG) is appropriate.

(c) Software installer — lossless. A program file must be reconstructed bit-for-bit; a single changed byte could stop the software running, so the compression must be lossless (e.g. ZIP).

How RLE compresses a bitmap: RLE scans the pixels (e.g. row by row) and replaces each 'run' of consecutive identical pixels with a single pair: the number of pixels in the run followed by the colour value. For example, 20 identical white pixels are stored as one pair (20, white) instead of 20 separate values. Where an image has large areas of the same colour, many long runs are replaced by short pairs, greatly reducing the file size. The process is lossless — every original pixel can be reconstructed exactly.

Why a photograph may not shrink: A photograph has continuously varying colours, so consecutive pixels are rarely identical and most runs have length 1. RLE then stores each single pixel as a count (1) plus its value — that is more data than the pixel by itself. As a result the 'compressed' file can be the same size or even larger than the original.