Video face detection, background insertion, automatic camera settings, and more (DEAL 389)

This portfolio is generally related to image processing. Various techniques are disclosed to process video images data based on presenters persona such as upper torso, shoulders, arms, hands, classifying pixels as foreground using both short-range depth data and long-range depth data and generate a virtual camera viewpoint for an image. Disclosed is a technique to receive video data frames of an ongoing scene and identify a persona. The system identifies a persona in the received frames by classifying a set of pixels which represents the persona, selects brightness values and adjusts video capture settings. Also disclosed are techniques to output the selected adjustments commands to the video camera and continue capturing the data of the scene. Few patents in the lot disclose a technique to remove background information from a video by categorizing pixels as foreground (FG) or unclear (UC) by using colour and background history information to categorize UC pixels as FG or remove pixels and use FG pixels to create video with required background. A patent in the lot discloses a technique to create user personas by combining probability maps, visually de-emphasize displayed persona by identifying user focus and transmit 3D images using depth images. The technology may be implemented in digital video cameras, DSLRs, handy cams, mobile phones, video conferencing equipment, image processing software, online chat services, video conferencing services etc.