Abstract: An encoding method comprises receiving a plurality of video frames including first and third reference frames (F1, F3) and a target frame (F2); generating a temporary frame based on assumed constant-velocity motion between the first and third reference frames (F1, F3); comparing the temporary frame target frame to detect nonlinear motion (errors in the temporary frame) between the first and third reference frames and to produce motion vector information between the target frame and at least one of the reference frames according to the comparison; and encoding the reference frames and the motion vector information.

Abstract: A value of one or more Lagrangian multipliers is adaptively estimate and update based, at least in part, on the video source statistics or dynamic programming. Methods, techniques, and systems involve determining a first Lagrangian multiplier with a video encoder, and updating a second Lagrangian multiplier with the first Lagrangian multiplier. The system can include a Lagrangian multiplier Estimation Module that estimates the Lagrangian multiplier, and a Lagrangian multiplier Update Module that updates the current Lagrangian multiplier using the estimated Lagrangian multiplier. The Online Lagrangian Multiplier Estimation Module may function with Rate Distortion Slope Estimation with Rate Distortion Optimized Mode Decision; Rate Distortion Slope Estimation with Local Approximation; Rate Distortion Slope Estimation with Local Information; or Rate Distortion Slope Estimation with Global Information.

Abstract: In a system installed in a vehicle, a pickup unit repeatedly picks up a frame image of a target region including a road ahead of the vehicle, and an extracting unit extracts an edge of a lane marker formed on the road from each of the frame images. A creating unit creates a history of the edges of the lane marker for the respective frame images. A detecting unit detects pitch of the vehicle. The creating unit omits, as a target edge, an edge from the history of the edges for the respective frame images. The target edge to be omitted is extracted from a frame image that is picked up by the pickup unit with the pitch being detected by the detecting unit. A recognizing unit recognizes a region of the lane marker based on the history of the edges of the lane marker.

Abstract: A method for composing a three dimensional image with a long focal length and a three dimensional imaging system are provided. First, a zoom lens is adjusted to a first focal length to capture a first image and a fixed focal lens is used to capture a second image. Next, the zoom lens is adjusted to a second focal length to capture a third image. Then, a region occupied by the third image in the first image is positioned, a first disparity map between the first and second image in the region is calculated and magnified to obtain a second disparity map. Finally, a fourth image simulating the image captured by the fixed focal lens with the second focal length is composed by using the third image and the second disparity map, and the third and fourth image are output as a three dimensional image.