Practical Fields of Application

SHV Details

Simulated Human Vision..... Ian Overington

Location: Eastbourne. UK
ianoverington@simulatedvision.co.uk ............ www.simulatedvision.co.uk

About SHV

The extensive studies into the multiparametric trends of visual detection & recognition thresholds carried out in the late 1960’s & early 1970’s inevitably involved substantial explorations into the physiology & neuro-physiology of human vision. From this it was realised, by the mid 1970’s, that the whole of the interactive performance of human vision was, of necessity, accomplished by a single ‘pipeline’ form of processing (except for stereopsis, which operates with two identical pipelines with physically disparate viewpoints). It was found possible to demonstrate on early computers how several wide ranging & apparently disparate visual functions could indeed be carried out relatively simply based on a single input structure. This led to a computer vision schema which could be unified over a wide variety of tasks (VISIVE). At the same time it was possible largely to validate the earlier multiparametric vision modelling.

Eventually this resulted in the publication of a textbook (‘Computer Vision: a unified, biologically-inspired approach’) which covered the whole (rather vast) subject in a progressive manner, starting with a summary of the basic visual neuro-physiology and building from single frame basic edge detection, through paired-frame edge detection inter-related with local disparity measurement (for either temporally or spatially disparate frames - i.e., motion or stereo), to edge-driven region segmentation, local contour analysis, image reconstruction & texture disruption boundary determination.