The purpose of this research is to predict a bulk physical property of a cotton sample using digital image processing on a video image taken of a surface of the sample. Gravimetric bark content was chosen as the bulk physical property to use. Classical digital image processing techniques are used on monochrome video images of cotton samples to isolate trash areas. Geometric properties (area, perimeter, and equivalent ellipse major and minor axes) are used to predict the gravimetric bark content in 47 cotton samples with varying bark and total trash content.