There are two basic ways to improve the speed of a low-level vision algorithm: careful code optimization and selective processing. Reducing the computational effort expended on each pixel reduces the time required to process an image by a constant factor. Selective processing on a limited portion of an image using a focus of attention can decrease overall computation by orders of magnitude. A fast pixel-based algorithm is developed that uses these principles to achieve fast extraction of lines for use in vision-guided mobile robot navigation. It builds upon an algorithm for extracting lines by grouping pixels with similar gradient orientation. It allows parametric control of computational resources required to extract lines with particular characteristics.>

The author seeks to exploit the inherent continuity of visual motion by examining the trajectory of a moving edge in densely sampled imagery. Past edge motion is propagated ahead of the moving edge along the trajectory; this information is then used to evaluate current motion measurements and compute important feature properties. Two related temporal image features are presented. Temporal persistence is the period of time over which a spatial feature has stably persisted in the image plane. Consistent edge motion occurs when an edge moves in a manner which is consistent with its previous trajectory. These temporal features are important to object recognition because spatial structures which persist in a stable manner over time are likely to be structurally related to objects. Results on natural imagery are presented.>

This paper proposes an intensity-based method for determining the spatial orientation and observed velocity of a continuously moving edge. Space-time relations arise from the translation of a straight edge across a fixed detector array; a tesselation unit comprised of three equidistant detectors is used to exploit these relationships. The time ordering of cells within this unit can estimate edge direction and velocity. More precise edge determination is possible when intercell times are available and constant edge velocity is assumed; a sensitivity analysis examines the effect of relaxing the velocity constraint. The simulation of a translating synthetic image over an array of detectors is presented in order to demonstrate the method.