Introduction

You can train Media Server to recognize objects, such as logos, that appear in images and video. These objects can be two-dimensional or three-dimensional.

2-D object with 2-D similarity transform

For example, a company logo on a scanned document. Media Server can still recognize the logo when it is subject to 2-D similarity transformations such as rotation and scaling.

2-D object with perspective transform

For example, a photograph or video frame that shows a sign displaying a company logo.

Perspective transformations result from viewing the object from any angle other than directly perpendicular to its plane. This can result in skewing of the image. Perspective transformations often occur in photographs of objects when the camera axis is not directly aligned with the object.

Media Server can only recognize 2-D objects that appear on flat, rigid objects. For example, if you attempt to recognize a logo printed on a flag or item of clothing, recogntion may not be reliable.

3-D object

For example, a photograph or video frame that shows product packaging.

If you supply sufficient training images, Media Server can recognize a 3-D object from any angle.


_HP_HTML5_bannerTitle.htm