We describe a method for identifying undesirable items in a camera-captured document that should be removed (e. g., one or more fingers holding the document, coffee stains, smudges on paper, etc.) during the process of automatically retouching and enhancing the image of the captured document. In the proposed system, an image processing system is responsible for identifying regions of interest within an image, which are then sent to an image classification module for further recognition and possible removal. This module may be a Machine Learning-based classifier, and is responsible for accurately identifying if a removable object is contained within the region. A previously published system  describes the overall object recognition and removal workflow, and here we focus instead on a specific approach for selecting candidate artifacts for removal.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 License.
INC, HP, "METHOD FOR RECOGNIZING AND REMOVING UNWANTED MARKS FROM SCANNED DOCUMENTS IMAGES USING SPECIALIZED IMAGE CLASSIFIERS", Technical Disclosure Commons, (September 15, 2022)