Video storage

What is this patent about?

'969 is related to the field of video encoding, specifically techniques for adjusting the quantization parameter (QP) during video compression. The background involves standard video coding methods like MPEG and H.264, which compress video by removing redundant temporal and spatial information. A key challenge is selecting the appropriate QP to balance bit rate and video quality, especially when hardware limitations prevent looking ahead at future frames to predict motion.

The underlying idea behind '969 is to use motion-sensing data from accelerometers or location-sensing circuitry as a proxy for predicting motion in video frames. Instead of analyzing future frames, the invention leverages the device's physical movement to estimate the amount of motion that will be present in the captured video. This allows for dynamic adjustment of the QP, improving video quality without requiring extensive buffering or look-ahead capabilities.

The claims of '969 focus on an electronic device (e.g., a smartphone) with a camera, motion-sensing circuitry, and data processing circuitry. The device captures an uncompressed video frame, detects physical motion, and encodes the frame using a QP. The QP is selected based on the detected physical motion and a prediction error of a preceding encoded frame, representing a difference between the encoded frame and a source frame from which the encoded frame was coded.

In practice, the device captures video and simultaneously monitors its own movement using accelerometers or location sensors. If the device detects significant movement, it lowers the QP to preserve detail in the encoded video, anticipating that the movement will introduce more complexity. Conversely, if the device is relatively still, it increases the QP to reduce the bit rate, assuming less detail needs to be preserved. The system also compares the prediction error of previous frames with the motion-sensing input to determine if the motion data is actually indicative of motion in the video.

This approach differs from prior solutions that rely on analyzing future or past video frames to estimate motion. By using real-time motion-sensing data, the invention avoids the need for complex look-ahead algorithms or large frame buffers. This is particularly beneficial for devices with limited processing power or memory, enabling them to dynamically adjust the QP and achieve a better balance between video quality and compression efficiency. The correlation between prediction error and motion-sensing input is crucial for determining the QP.