System and method for script and orientation detection of images using artificial neural networks,申请号US201213442892-传众专利搜索

System and method for script and orientation detection of images using artificial neural networks

申请公布号：US8891822(B2)

申请号：US201213442892

申请日期：2012.04.10

申请公布日期：2014.11.18

申请人：

Hewlett-Packard Development Company, L.P.

发明人：Jain Chirag;Goudar Chanaveeragouda Virupaxgouda;Srinidhi Kadagattur Gopinatha;Wu Yifeng

分类号：G06K9/00;H04N5/228

主分类号：G06K9/00

地址：Houston TX US

摘要：A system and method for script and orientation detection of images using artificial neural networks (ANNs) are disclosed. In one example, textual content in the image is extracted. Further, a vertical component run (VCR) and horizontal component run (HCR) are obtained by vectorizing each connected component in the extracted textual content. Furthermore, a zonal density run (ZDR) is obtained for each connected component in the extracted textual content. In addition, a concatenated vertical document vector (VDV), horizontal document vector (HDV), and zonal density vector (ZDV) is computed by normalizing the obtained VCR, HCR, and ZDR, respectively, for each connected component. Moreover, the script in the image is determined using a script detection ANN module and the concatenated VDV, HDV, and ZDV of the image. Also, the orientation of the image is determined using an orientation detection ANN module and the concatenated VDV, HDV, and ZDV of the image.

主权项：1. A method for script and orientation detection of an image using artificial neural networks (ANNs), comprising: extracting textual content in the image; obtaining a vertical component run (VCR) by vectorizing each connected component in the extracted textual content into a plurality of horizontal zones and determining a number of vertical cuts in each of the plurality of horizontal zones for each connected component in the extracted textual component; obtaining a horizontal component run (HCR) by vectorizing each connected component in the extracted textual content into a plurality of vertical zones and determining a number of horizontal cuts in each of the plurality of vertical zones for each connected component in the extracted textual component; obtaining a zonal density run (ZDR) for each connected component in the extracted textual content; computing a concatenated vertical document vector (VDV), horizontal document vector (HDV), and zonal density vector (ZDV) by normalizing the obtained VCR, HCR, and ZDR, respectively, for each connected component in the image; determining the script in the image using a script detection ANN module and the computed concatenated VDV, HDV, and ZDV of the image; and determining the orientation of the image using an orientation detection ANN module and the computed concatenated VDV, HDV, and ZDV of the image.