Multimedia Database Indexing & Search
PixSearch enables content-based indexing, search, and retrieval of multimedia content in large databases. With a set of unique appearance-based queries – images, objects, and videos can be easily retrieved based on visual cues. This includes built-in features such as color, texture, and shape, as well as customized, domain-specific features and metrics that can be easily integrated into the system.
To enable search of tracked objects in multi-camera environments, such as in surveillance systems deployed at airport terminals, cities, roads, and other large sites, PixSearch VIDINT (Video Intelligence) uses additional data, including object properties, track segments, events, and site topology. PixSearch VIDINT enables security personnel and law enforcement officials to “get the big picture” by gaining overall site-wide understanding and enhancing of video intelligence capabilities for debriefing, forensic analysis, and crime prevention.
The PixSearch technology family is divided into two main categories:
- PixSearch Indexing Engine, a multimedia content-based indexing and search engine
- PixSearch VIDINT (Video Intelligence), object-based, indexing and search for multi-camera environments
The PixSearch Indexing Engine provides a technology suite and API that can easily integrate into various products and applications, enabling users to quickly locate the most similar images and other multimedia objects according to their visual content, navigating through large multimedia databases to intuitively find results. Using PixSearch, queries can be performed by simply describing the desired image content or by providing an image sample as the query criteria. Similarity, in this respect, is based on a set of features that can be either automatically extracted from the image, such as color, texture, and shape, or provided by the application by defining domain-specific features and metrics.
The Indexing Engine incorporates cutting edge technologies. In the indexing process, appropriate signatures are extracted from the images/videos and are efficiently indexed for fast and easy search and retrieval based on perceptual metrics. Designed for high performance with very large databases, the unique database query technology and advanced data structures enable efficient multi-modal search, utilizing standard SQL-based commercial databases. It also enables advanced search by description, utilizing a sophisticated query language.
The PixSearch Indexing Engine is particularly attractive for companies in need of indexing and content-based search capabilities for a large collection of objects (e.g., hundreds of millions) where traditional object tagging and labeling is not possible or inefficient. If a similarity can be defined between two objects in the collection, the PixSearch Indexing Engine automatically provides efficient indexing and the ability for fast search, including query-by-example and query-by-description.