A parallel Markov variational optical flow determination method and system

Abstract

The invention discloses a parallel Markov variational optical flow determination method and system. The method includes: acquiring any two consecutive frames of images in real time; calculating multiple data items in the Markov optical flow model in parallel according to the two frames of images; calculating the information to be transmitted by each image pixel according to the data items, according to The information iteratively updates the message of each pixel in parallel to obtain the updated message; at the same time, the updated message is transmitted in parallel to the four directions of up, down, left and right of each image pixel; whether the number of transmissions is judged The iteration threshold is reached; if yes, output a transfer message, wherein the minimum information in the transfer message corresponds to the optical flow value to be calculated; if not, jump to the update message calculation step. The method for determining the parallel Markov variational optical flow involved in the present invention adopts a multi-thread structure to calculate and transmit information in four directions in parallel, up, down, left, and right, and greatly improves the calculation efficiency.

Description

technical field [0001] The invention relates to the field of optical flow calculation, in particular to a parallel Markov variational optical flow determination method and system. Background technique [0002] Image sequence variational optical flow calculation technology has made some progress in the detection of moving objects, and has also been applied and played an important role in social production, life and other fields, but the disadvantage is that the time consumption of variational optical flow calculation is usually large, and it is difficult to meet the target. Real-time applications such as detection. The conventional parallel BP algorithm simply parallelizes the serial calculation steps by simply taking each pixel in the image as a unit. It usually uses a two-dimensional parallel computing framework, which is not optimized for the characteristics of BP calculation. Due to information calculation Four directions are involved, and the information calculation in ...

AI Technical Summary

Problems solved by technology

[0002] Image sequence variational optical flow calculation technology has made some progress in the detection of moving objects, and has also been applied and played an important role in social production, life and other fields, but the disadvantage is that the time consumption of variational optical flow calculation is usually large, and it is difficult to meet the target. Real-time applications such as detection

The conventional parallel BP algorithm simply parallelizes the serial calculation steps by simply taking each pixel in the image as a unit. It usually uses a two-dimensional parallel computing framework, which is not optimized for the characteristics of BP calculation. Due to information calculation Four directions are involved, and the calculation of information in each direction must be calculated using a complex dimensionality reduction method, resulting in a single-thread calculation load that is too high and calculation efficiency is not high

Images