Session Track:

Visual Harmonization Metric of the Multi-camera Visualization in Automotive Surround View Systems



In existing multi-camera automotive vision systems, many different 2D and 3D views (curbview, U-view, Surround View, Bowlview) can be created utilising 2, 3 or 4 cameras. Each camera independently adjusts certain Image Signal Processing (ISP) parameters to obtain the best image for its own field of view. Examples of this would be exposure control and white balancing. When merged together in a multi-camera view (e.g. surround view), it is very obvious when the road surface in the view has a different brightness and colour from each camera. Harmonization is a post processing image quality algorithm that balances the brightness and colour of pixels from each camera used to create these multi-camera views. This metric is a proposed objective measurement to quantify the visual quality performance of the surround view with respect to harmonization.

Hear from:


Mark Griffin
Image Quality Algorithm Developer,



No data was found
Scroll to Top