When a surface slanted away from the fronto-parallel plane is viewed binocularly, surface markings and texture are imaged with slightly different orientations and degrees of foreshortening. These orientation and spatial frequency disparities are systematically related to surface slant and tilt and could potentially be exploited by biological and machine vision systems. Indeed, there is evidence suggesting that human stereopsis has a mechanism that specifically makes use of orientation and spatial frequency disparities, in addition to the usual cue of horizontal positional disparity. In machine vision algorithms, orientation and spatial frequency disparities are a source of error in finding corresponding points in left and right views, because one seeks to find features (or areas) which are similar in the two views when, in fact, they are systematically different. In other words, it is common to treat as noise what is useful signal.
We have been developing a new stereo algorithm based on the outputs of linear spatial filters at a range of orientations and scales. We present a method in this framework for making use of orientation and spatial frequency disparities to directly recover local surface slant. An implementation of this method has been tested on curved surfaces and quantitative experiments show that accurate surface orientation can be recovered efficiently. This method does not require the explicit identification of oriented line elements and also provides an explanation of the intriguing perception of surface slant in the presence of orientation or spatial frequency disparities, but in the absence of systematic positional correspondence.