System and method for script and orientation detection of images

摘要

A system and method for script and orientation detection of images 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 concatenated vertical document vectors (VDV) and a horizontal document vector (HDV) are computed. In addition, a substantially matching script and orientation is obtained by comparing the computed concatenated VDV and HDV of the image with reference VDV and HDV associated with each script and orientation, respectively. Also, the substantially matching script and orientation are declared as the script and orientation of the image, if the computed concatenated VDV and HDV of the image substantially match with the reference VDV and HDV of the matching script and orientation, respectively.

主权项

1. A method for script and orientation detection of an image, comprising:
extracting textual content in the image; obtaining a vertical component run (VCR) and a horizontal component run (HCR) by vectorizing each connected component in the extracted textual content in the image; computing a concatenated vertical document vector (VDV) and horizontal document vector (HDV) by averaging the obtained VCR and HCR for each connected component in the image; obtaining a substantially matching script and orientation by comparing the computed concatenated VDV and HDV of the image with each of a plurality of reference VDVs and HDVs, wherein each reference VDV and HDV is associated with a script and an orientation of a plurality of scripts and orientations; determining whether the computed concatenated VDV and HDV of the image substantially match with the reference VDV and HDV associated with the matching script and orientation; and if the computed concatenated VDV and HDV of the image substantially match with the reference VDV and HDV associated with the matching script and orientation, then declaring the matching script and orientation as the script and orientation of the image.