Here three histograms of the stereo disparities of detected features are shown. The red lines show an estimate of the peak histogram value. When viewing more distant objects the histogram clusters towards the left (one or two pixels of disparity). Nearer objects tend to produce a wider span of disparity values. There is always a certain amount of noise, due to features existing at different depths, noise within the image, and erroneous stereo matches. For more info see http://code.google.com/p/sentience/
Link to this comment:
All Comments (0)