Dismiss
The Prior Art Database and Publishing service will be updated on Sunday, February 25th, from 1-3pm ET. You may experience brief service interruptions during that time.
Browse Prior Art Database

# Method for SIMD checkerboard block matching for video motion estimation

IP.com Disclosure Number: IPCOM000023824D
Publication Date: 2004-Mar-31
Document File: 2 page(s) / 30K

## Publishing Venue

The IP.com Prior Art Database

## Abstract

Disclosed is a method for single instruction multiple data (SIMD) checkerboard block matching for video motion estimation. Benefits include improved performance.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 63% of the total text.

Method for SIMD checkerboard block matching for video motion estimation

Disclosed is a method for single instruction multiple data (SIMD) checkerboard block matching for video motion estimation. Benefits include improved performance.

Background

Motion estimation involves taking a block (typically 16x16 pixels) from the luminance plane of video frame N in a video sequence. The block in video frame N-1 that most closely approximates the block in frame N is identified. The identified block is encoded by transmitting the location of the predictor block in frame N-1 and some fix-up pixel values to compensate for the fact that the predictor block is only an approximation.

The core component of motion estimation is a function that compares two blocks to see how similar (close) they are (see Figure 1). The values x and y are the two blocks being compared. The sum is over all of the 16x16 pixels.

Because motion estimation is a very time-consuming process, many researchers have developed ways of reducing the amount of computation required. One technique involves calculating the sum over a subset of the 16x16 block.

Another popular method that works well is to calculate the sum over just the black squares of a checkerboard (the pixels, see Figure 2). This solution cuts the work in half with little effect on the quality of the match that results. Subsampling by more than 1/2 (such as 1/4) is more problematic. The quality of the match starts to suffer significantly.

The problem with thes...