摘要 |
A method for pre-filtering visual objects on a document is disclosed. A selection of a template visual object with constituent components from a subsection of the document is received. Then, a feature set is derived, including a color list of colors defining the object components, a bounding definition of the object components, and an ancillary raster component intersection flag. A subset of pre-filtered visual objects is generated from the visual objects on the document. These match a predefined criteria based on the color list, the bounding definition, or the ancillary raster component intersection flag. A pre-filtered document raster image is then generated from the document, where the pixels corresponding to the pre-filtered visual objects are included. |