Text-to-document image classifier

Requirements Background: File management in various industries is an important work, with the passage of time, a variety of paper files more and more information, file archiving, preservation, access to work is increasingly difficult, the industry in the archives management, file re-use of the cost of more and more. With the development of it technology, the management of archival image (electronic) has been accepted by more and more users, through the construction of Archives image (electronic) management system, users can achieve the following goals: After the image of the archives, the original files can be sealed and stored, and the electronic access and electronic management of the archives can be realized by indexing the images. After the electronic file can be stored for a long time, the original files can be destroyed after the legal period of retention, saving storage, custody, transportation and other costs; The electronic file has improved the efficiency of archival access, thus improving the work efficiency and reducing the labor cost. At present, in the process of archival electronic, can not achieve automatic classification and management of file image, that is, the computer can not automatically determine whether an image is an invoice, order or express, Image classification archive is the use of manual classification and business process Control completed, low level of automation, the need to improve the efficiency of classification. Text-to-document image classifier is used. Product Overview: Document Image classifier is a kind of automatic classification of images, which can not only complete the automatic classification of various instruments in financial risk supervision system, but also can realize automatic classification of other forms and archives images, such as express orders, invoices, vouchers and so on, which can be intelligently distinguished from different courier companies. The classifier consists of two parts: the template editor and the automatic classification core. The template editor completes the customization of the template, and the classification template directs the classifier to classify the image. Product Features: 1, image preprocessing: to Black edge, to bookbinding hole, tilt correction and so on; 2, classification template Editing tools: script editing tools; 3, classifier Engine: Image automatic classification, archiving output; Product Features: 1, the template classification of fine, accurate rate high The accuracy of template classification is more than 99%; classifiers support simultaneously Table frame Line (position, color) classification; Title Unit (title content, title color) classification; Text area (text content, text color) classification; Image feature extraction classification; Classifier to maximize the difference between different vouchers. such as the border color, the title color can be used to distinguish the multi-linked vouchers, the text content can be distinguished from the same document to print multiple transactions of the image; The text color can be used to distinguish between the same voucher and the same color of the border between the lines and the title. 2. Fast classification speed After the actual production line operation calculation, the single voucher classification time is 100ms~200ms. Our services are: In combination with your business scenarios, recommend excellent image capture, classification and management solutions; Integrated development technology Support, in the Enterprise Customer development application system process, we track all the way, to provide customers with professional team of technical services; To provide customers with continuous software upgrade services; Application areas/Solutions: Express Single sorting

Text-to-document image classifier