Compression of Bayer Colour Filter Array Images

Abstract

Bayer Colour Filter Array is a matrix of photosensors covered with red, green, and blue colour filters. This setup is advantageous in smartphones as only a third of the required data is captured by the sensor in the camera. The rest of the components based on the colour format can be interpolated using a suitable algorithm to arrive at a full-colour image. Increasing the resolution of the camera sensor will translate to increased bandwidth in the image signal processing pipeline, and consequently power consumption. In addition to that, the bit depth is also on the rise to enhance the colours. These two factors will create a huge impact on the data to be handled in the pertinent processor. Hence, compression of the Bayer data is of immense significance. The existing standard compression schemes can be adapted to suit the Bayer format. Also, several compression schemes, specific to Bayer format have been proposed. Two compression methods, viz. JPEG-LS and Hierarchical Prediction based compression have been tested and the corresponding results are presented in this paper. The former is a standard while the latter has been proposed keeping the Bayer format in mind. Modelling of the algorithms shows that JPEG-LS is best suited in the use cases where lossless compression is desirable, and Hierarchical Prediction based compression is the better option where some amount of loss is acceptable.