This thesis deals with focus of attention control in active vision systems. A framework for hierarchical gaze control in a robot vision system is presented, and an implementation for a simulated robot is described. The robot is equipped with a heterogeneously sampled imaging system, a fovea, resembling the spatially varying resolution of a human retina. The relation between foveas and multiresolution image processing as well as implications for image operations are discussed.
A stereo algorithm based on local phase differences is presented both as a stand alone algorithm and as a part of a robot vergence control system. The algorithm is fast and can handle large disparities and maintaining subpixel accuracy. The method produces robust and accurate estimates of displacement on synthetic as well as real life stereo images. Disparity filter design is discussed and a number of filters are tested, e.g. Gabor filters and lognorm quadrature filters. A design method for disparity filters having precisely one phase cycle is also presented.
A theory for sequentially defined data modified focus of attention is presented. The theory is applied to a preattentive gaze control system consisting of three cooperating control strategies. The first is an object finder that uses circular symmetries as indications for possible object and directs the fixation point accordingly. The second is an edge tracker that makes the fixation point follow structures in the scene. The third is a camera vergence control system which assures that both eyes are fixating on the same point. The coordination between the strategies is handled using potential fields in the robot parameter space.
Finally, a new focus of attention method for disregarding filter responses from already modelled structures is presented. The method is based on a filtering method, normalized convolution, originally developed for filtering incomplete and uncertain data. By setting the certainty of the input data to zero in areas of known or predicted signals, a purposive removal of operator responses can be obtained. On succeeding levels, image features from these areas become 'invisible' and consequently do not attract the attention of the system. This technique also allows the system to effectively explore new events. By cancelling known, or modeled, signals the attention of the system is shifted to new events not yet described.
Linköping: Linköping University Electronic Press , 1995. , 185 p.
Knutsson, Hans, Dr.