b'Webwaves Webwavescosine transform (DCT), which was first published by Ahmed, Natarajan & Rao in 1974. The DCT uses a sum of different frequency cosine functions to express a finite sequence of data points. To those of us used to looking at seismic data, a good analogy is Fourier transforms and their uses. The JPEG image format uses DCT in a similar approach to resampling seismic data, with frequencies of interest retained and other frequencies rejected. Other lossy approaches include using discrete wavelet transforms (DWT) and Ian Jamesmodified discrete cosine transforms ASEG Webmaster(MDCT).webmaster@aseg.org.auLossy compression is widely used both on the internet and on local machines. Multimedia files are commonly Lossless vs lossycompressed with lossy compression compression algorithms, and file formats like JPEG, MP3, MP4 and MKV all represent files In the high bandwidth world, we findthat have used a lossy algorithm.Figure 1.a) Image from the 2018 photo ourselves in today, you could be forgivenThe benefit of lossy algorithms iscompetition, and featured on the Preview page of that, when used judiciously, minimalthe website. Compressed as JPEG, file size of 116 KB. if you ignored data compression.degradation of the end product willMinimal loss in the image is observed at this scale. However, with video streaming andoccur, while being able to encode theb) Further compressed version of a). Format JPEG, social media viewing consuming internetdata in a significantly smaller file. Thisfile size 20 KB. Considerable loss of detail is observed traffic, data compression is just ashas obvious benefits on the internet,by the additional compression.important today as it was in the dial-upresulting in faster loading times, and days of yesteryear. reduced network requirements. In thisIn contrast, conversion between There are two commonly usedrespect, some loss may be desirable tolossless formats performs well as terms in data compression: losslessimprove performance for the end user. the data is able to be reconstructed compression and lossy compression.When it comes to lossless compression,perfectly within the compression. For Lossless compression uses a group oftext files are useful to consider. In aexample, when the Microsoft Office algorithms that allows the original datastandard ASCII file, an 8-bit code issuite moved to the .*x file extensions to be accurately reconstructed from theassigned to each character, resulting in(e.g. .docx, .pptx), the files became compressed. Lossy algorithms do nota constant and expected file size basedcompressed as a zip archive. As a facilitate accurate reconstruction of theon the number of characters. Contrastresult, compression of Microsoft Office original data and there is some loss ofthis with Morse code, where commondocuments offers little advantage in information in the compression. letters have fewer dots and dashes toterms of file size.With compression, there is no rightbe encoded. This naturally leads us toOn the ASEG website we use a variety answer for how best to compress theHuffman coding, whereby symbolsof lossless and lossy compression data, with advantages and disadvantagesare encoded based on the frequencyalgorithms in our media. With high of both lossy and lossless approaches.of occurrence, with higher frequencyquality imagery provided via the photo An approach can be considered basedsymbols having shorter sequences, as incompetition, some images are displayed on the acceptable quality and size of theMorse code. at high resolution using lossless file output data. For instance, on the web,formats, while other, non-critical priority may be given to lower resolution,Should you have data in a lossycontent,is displayed using highly lossy data formats to acceleratedata format, converting the datacompressed lossy formats (seeFigure1). viewing. But repeated compressionto a lossless format offers limitedVideos hosted via YouTube, such as anddecompression can result ingains, with no real benefit overthe talks that are published online, information loss. keeping the data in the lossy format.will also be compressed using Similarly, conversion between lossyGoogles algorithms, with users able to One of the most widely used lossyformats results in increased loss ofchoose their desired resolution when compression algorithms is the discretedata if different algorithms are used.streamingthem.APRIL 2020 PREVIEW 42'