Visual Evoked Potentials for Image Quality Assessment with Scale Space Approach

This paper presents an objective assessment method of image quality using visual evoked potentials (VEPs) to image engineer field based on multi-diseiplinarily approach such as knowledge of neurobiology, image recognition theory of computer vision. The multi-disciplinarily based objective assessment method applies Gaussian scale-space filtering in order to consist a scalar parameter to depict blur image. In experiment visual stimuli are provided the scalar parameter and subjects are detected VEPs. Their VEPs are recoded during observation of the checkerboard pattern reversal (PR) stimuli, and are analyzed latency of about Negative 145 msec (N145) component. The result of the experiment was that latency of N145 components were long about10-20 msec when parameters were large vale (more blur). This result shows one example of availableness for the multi-disciplinarily based objective assessment of image quality by integrating the pattern reversal visual evoked potential (PR-VEP) and the scale-space theory.