Video trashmeters are currently used in cotton classing offices and textile mills to predict cotton grade from the number of trash pixels and the number of cotton-to-trash gray level transitions. In these systems, a video image is digitized into a rectangular array of picture elements (pixels). Each pixel contains a value of "whiteness" (0-255). Trash is typically darker than cotton, so pixels below a threshold are declared trash. A limitation of these systems is their inability to distinguish between trash categories such as leaf trash, seedcoat fragments, or pieces of bark or grass. In this research, we measure the shape and other parameters for each trash object. Results of this initial study have shown it is possible to categorize trash into "bark/grass" versus "other" by utilizing trash shape, perimeter, area, and circumscribed box area measurements.