Improving ‘Objective’ Digital Images with Neuronal Processing: A Computational Approach

• Published in: International journal of architectural computing 2004-06, Vol.2 (2), p.143-153
• Main Authors: Mahalingam, Ganapathy, Kavasseri, Rajesh G.
• Format: Article
• Language: English
• Subjects: Balancing; Charge coupled devices; Computation; Digital cameras; Luminosity; Mathematical models; Rendering
• ISSN: 1478-0771; 2048-3988
• DOI: 10.1260/1478077041518692

This paper describes an experiment where an image recorded with a digital camera is processed using an electro-physiological model of a neuron. The luminosity level of each pixel of the source image is treated as the stimulus for an individual neuron, and the source image is transformed into a response image based on the processing behavior of the Hodgkin-Huxley neuronal model. It is seen that transformation of the image through neuronal processing yields (i) more evenly balanced levels of luminosity and (ii) a more ‘subjective’ rendering of the environment than what was photographed with the digital camera. The CCD (charge coupled device) – based digital camera reveals its limitation as a linear recording device that does not have a balanced dynamic range. The neuronal processing of the image adds non-linearity and a balanced range to the luminosity levels in the image, rendering it closer to a ‘subjective’ perception of the scene.