Stretching and warping said to provide better data compression
1 min read
Researchers from UCLA have created a new data compression method that is said to outperform existing techniques such as Jpeg. The team says the approach could eventually be adopted for medical, scientific and video streaming applications.
The work has been undertaken in an attempt to keep up with the growing number of applications which require massive amounts of data to be captured and analysed in real time.
The group, led by Professor Bahram Jalali, has drawn on the graphic art technique of anamorphism. The group discovered that it is possible to compress data by stretching and warping it using a newly developed mathematical function. The technology – anamorphic stretch transform, or AST – operates in the analogue and digital domains. In analogue applications, AST makes it possible to capture and digitise signals that are faster than the speed of the sensor and the digitiser and also to minimise the volume of data generated in the process.
For digital records – for example, medical data – the transformation causes the signal to be reshaped such that 'sharp' features are stretched more than 'coarse' features.
According to the research team, the technique does not require prior knowledge of the data for the transformation to take place; it occurs naturally and in a streaming fashion.
Postdoctoral researcher Mohammad Asghari said: "Our transformation causes feature selective stretching of the data and allocation of more pixels to sharper features, where they are needed the most. For example, if we used the technique to take a picture of a boat on the ocean, AST would cause the sailboat's features to be stretched much more than the ocean, to identify the boat while using a small file size."
Tests have shown that AST can outperform standard Jpeg image compression, with a 'dramatic' improvement in terms of image quality and compression factor.
AST is said to have its origin in another technology pioneered by Prof Jalali's group; the time stretch dispersive Fourier transform. This is a way to slow and amplify faint, but very fast, signals so they can be detected and digitised in real time.
According to Prof Jalali: "Reshaping the data by stretching and wrapping it in the prescribed manner compresses it without losing pertinent information. It emulates what happens to waves as they travel through physical media with specific properties. It also brings to mind aspects of surrealism and the optical effects of anamorphism."
