next up previous
Next: Strategy and Planning Up: Vision Processing Previous: Symmetry Detection

Color Model Classification

We have now localized a set of possible candidates for objects of interest (balls and pockets). To recognize and classify these objects, we propose the use of color models. We train a set of probabilistic color models for the pockets and the balls (cue ball, 8-ball, the seven striped balls and the seven solid balls). We also train models of other objects that might appear in the image as false alarms: a color model of the pool cue and the player's hand. This process is identical to the color modeling used for the table and we obtain a total of 20 models which have a form similar to Equation gif. We shall refer to these models as tex2html_wrap_inline365.

Figure gif and Figure gif show the color modeling process for the solid red ball and the striped orange ball. For each model, we begin with a distribution of pixels in RGB space as shown in Figure gif(a) and Figure gif(a). These sets of 3 dimensional RGB vectors are labeled tex2html_wrap_inline367. For each tex2html_wrap_inline369 we compute a model tex2html_wrap_inline371 when we first train the system.

  figure202
Figure: The Solid Red Ball (Trained)

  figure211
Figure: The Striped Orange Ball (Trained)

Next we apply all 20 color models to examine each of the possible candidate symmetry peaks. Around each peak, we collect a small window of pixels. All non-table pixels (i.e. the ones that did not match the table's color model) are collected to form a distribution of RGB data similar to the one in Figure gif(a). This test distribution will be called tex2html_wrap_inline373. We form a single Gaussian color model (because EM with multiple Gaussians would be too slow to compute online). This model, shown in Figure gif(b), is called tex2html_wrap_inline375.

  figure223
Figure: The Test Object (Orange Striped Ball)

To classify our test distribution tex2html_wrap_inline373 we use a common distance metric between probabilistic models. This metric is the Kullback-Liebler divergence [4] gif. By measuring the 'distance' between test data tex2html_wrap_inline375 and our training data, tex2html_wrap_inline371 for tex2html_wrap_inline383 we can see how similar it is to other objects. We determine the closest model i for each symmetry peak and label that peak accordingly. This process is iterated over all the symmetry peaks which are ultimately labeled as solid ball, striped ball, cue ball, 8-ball, pocket and 'other'.


next up previous
Next: Strategy and Planning Up: Vision Processing Previous: Symmetry Detection

Tony Jebara
Wed Feb 18 18:52:15 EST 1998