Data Compression

Lossless Compression

• Run-Length Encoding (RLE): This method identifies sequences of identical data elements (runs) and replaces them with a single data value and a count. For example, a sequence of twenty identical white pixels in a simple graphic would be stored as the pair (20, White), significantly reducing the storage requirement for repetitive data.
• Huffman Coding: This technique uses a frequency-based binary tree to assign variable-length bit patterns to characters. Characters that appear most frequently in a file are given the shortest binary codes (e.g., 2 bits), while rare characters receive longer codes, ensuring the overall file size is minimized.

Lossy Compression

• Perceptual Encoding: This approach removes data that the human brain is unlikely to notice, such as sounds at frequencies beyond human hearing or subtle color variations in a complex photograph. By discarding this 'irrelevant' information, the file size can be reduced by 90% or more while maintaining a high perceived quality.
• Quantization: In image and audio compression, quantization reduces the precision of data values (like color depth or sound amplitude). While this introduces small errors or 'artifacts,' it allows the data to be represented with far fewer bits.

The choice between compression types depends entirely on the nature of the data and the consequences of losing information.

• Lossless is mandatory for files where every single bit is critical, such as executable programs or legal documents, where a single missing character could render the file useless. Lossy is preferred for media consumption where the human eye or ear can compensate for missing detail in exchange for faster loading times.

• Identify the Data Type: If an exam question asks which compression to use for a specific scenario, look for keywords like 'executable,' 'text,' or 'database' to choose lossless, or 'streaming,' 'web gallery,' or 'voice call' to choose lossy.

• Explain the 'Why': When asked to justify lossy compression, always mention that the discarded data is 'unnoticeable to humans' and that the primary benefit is 'reduced transmission time' or 'lower storage costs.'

• Huffman Tree Logic: Remember that in a Huffman tree, no code is a prefix of another code. This 'prefix property' is what allows the computer to decode variable-length bits without needing spaces between them.

• Check for Reversibility: A common exam trap is asking if a lossy file can be restored to its original state. The answer is always 'No'—once data is discarded in lossy compression, it is gone forever.