The piXsearch™ libraries analyze images and video frames, and create a searchable index of their contents.  The piXsearch libraries are able to “see” the visual objects in an image (such as a car, person, logo, etc.), and can compare images based on how similar these objects are. You can use the piXsearch API to add visual search capabilities to your application, including:

• Automatically index digital images and video files.
  Without manual intervention, the metadata needed to describe the contents of the image is extracted automatically by the piXsearch libraries.
  
• Find digital images and video segments based on visual content.
  By integrating piXsearch in your application, user will be able to retrieve the images/video segments they need by clicking and pointing on a picture.
  
• Find pictures and videos that contain specific text on the image/video.
  piXsearch makes searchable text that may be visible on the image itself.  Users can type a text string to retrieve images/video segments where that word appears on the screen.
  
• Integrate your application’s metadata with the visual contents of the image.
  The piXsearch generated metadata is easily accessible through XML in order to facilitate integration between data managed/stored by your application and the visual contents of the image.

iSome of the key advantages of the piXsearch SDK are:

• Accuracy-piXlogic's image segmentation, understanding, and search algorithms set a new benchmark of accuracy for the industry.
  
  
• Compactness- the footprint of the runtime environment is less than 3 Mbytes making it ideal to fit into existing applications,.  The core modules are even smaller in size,  which makes it suitable for implementation in memory limited environments such as handheld devices.
  
  
• Speed-it can scan millions of records in a fraction of a second and provide a prioritized list of matches.  The software is multithreaded providing linear scaling of performance with added CPUs, which allows for real-time analysis of video.
  
  
• Elegant Architecture-
  • Libraries are written in C++
    
  • Runs on the Windows platform (other OS ports upon request)
    
  • Includes Java Native interface
    
  • Thread safe API, enabling multiple indexing operations or database searches to be executed concurrently
  • API facilitates integration by providing simple and clean interfaces including:
    • Data stored in XML
    • Capability for using alternate tools to read various image/video file formats

Partial list of features:

Search Queries:

• Whole image search or object search within image.
• Composite searches of multiple objects from multiple pictures. Supports logical operations such as AND.
• Object recognition regardless of relative size, orientation and location within an image
• Supports nested object hierarchies (objects within an object)
• Combination of external metadata (for example keywords) and visual search metadata to formulate a search query
• Query images need not be part of an existing database.  All analysis and comparison calculations are done "on-the-fly".

User Preference Settings:

• Shape & Color-users can express a preference (eg, focus on objects that have the most similar shapes regardless of colors <--> focus on objects that have the most similar colors regardless of shape).
• Object-based search query formulation - users can interact with the query image to define their query visually and precisely.
• Results presentation - users can express whether the software should return a fixed number of images or only results that meet a user defined confidence level threshold.