EAGER: Automated Watermark and Moldmate Identification

鉴定古代（1750年以前）纸上的水印是确定其生产日期和来源的重要法医。对于纸上艺术品来说，水印是确定年代和鉴定的关键因素。即使对纸张分析专家来说，对水印进行分类和区分也是一项困难而耗时的任务。最近的研究表明，基于计算机的决策树可以大大减少水印识别和伴随识别同一模具制作的纸张的时间，同时增加结果的置信度。数字图像处理在这项任务中的应用是计算艺术史新领域的一个例子，它在过去十年中得到了爆炸式发展。通过这个项目，使用数字图像处理的自动化可以扩大水印和模式化识别的任务，以处理伦勃朗及其学生的印刷品和图纸中数百个单独的水印，并允许将这种识别扩展到大型博物馆的这些艺术品和小型博物馆和私人收藏中的数千个艺术品。预计这将激励这一宝贵的法医扩展到其他1750年以前的艺术家在纸上艺术品中使用的数千个平纹纸上的水印。这个项目在计算艺术史上的进步将为普通大众提供一条道路，让他们以一种新的、吸引人的方式参与艺术，而普通大众正逐渐适应计算机在日常生活中的使用。艺术史是文化遗产中最迟加入数字科学和人文学科的分支领域。该项目的成功将成为人文与科学之间进一步跨学科活动的主要推动力，从而减少两种文化的分离，这在日常生活各个方面的数字化以惊人的速度占据主导地位时至关重要。这个项目延续了计算艺术史这个新领域的开创性努力，它扩大了人们对数字图像处理作为艺术史研究中一个重要工具的接受程度，因为它极大地扩展了艺术史学家的知识范围。随着这一新兴领域的第一批工具被开发出来，越来越多的科学家和工程师被吸引到这些问题的挑战中，艺术史学家的具体需求无疑将揭示出需要在数字图像处理的基础科学方面取得进展的新任务。计算和视觉人文学科之间的这种联系也为消除工程科学和人文学科在各级教育中不断扩大的鸿沟提供了强有力的联系。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Identification of watermarks in antique (pre-1750) laid paper is a vital forensic in determining the date and origin of its production. For artworks on paper, watermarks are a key element in dating and authentication. Classifying and distinguishing watermarks is a difficult, time-consuming task even for paper analysis experts. Recent research has shown that a computer-based decision tree can reduce dramatically the time expended in the task of watermark identification and the concomitant identification of sheets of paper made on the same mold, while at the same time increasing the confidence in the result. The application of digital image processing to this task is one example of the new field of computational art history, which has exploded over the past decade. Through this project, automation using digital image processing can scale up the task of watermark and moldmate identification to handle the universe of the hundreds of individual watermarks in the papers of prints and drawings of Rembrandt and his pupils and allow expansion of this identification beyond these artworks in major museums to the thousands in smaller museums and private collections. This is expected to energize the expansion of this valuable forensic to the thousands of watermarks in laid paper used in artworks on paper by other pre-1750 artists. Advances in computational art history stemming from this project will offer a path for the general population, which is growing quite comfortable with the expansion of computer usage in everyday life, to engage with art in a new and appealing way. Art history is one subfield within cultural heritage that has delayed the longest in joining the emergence of the digital science and humanities. The success of this project will be a major enabler for further trans-disciplinary activities between the humanities and the sciences, thereby decreasing the separation of their two cultures, which is vital as the digitization of all aspects of daily life gains dominance at a scorching pace.This project continues the pioneering efforts in the new field of computational art history that is broadening acceptance of digital image processing as a vital tool in art history research as it tremendously expands the intellectual reach of art historians. As the first tools are developed in this nascent field and more scientists and engineers are drawn to the challenges of these problems, the specific needs of art historians will undoubtedly reveal new tasks requiring advances in the underlying science of digital image processing. This connection between computation and the visual humanities also offers a powerful link to combat the widening gulf in education at all levels between engineering science and the humanities.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.