Multi-view image processing method, device and electronic equipment including moving target

Abstract

The embodiment of the invention discloses a multi-visual angle image processing method and device including a moving target as well as electronic equipment, relating to the technical field of image processing. The method comprises the following steps: acquiring a first planar motion vector of the moving target in a first visual angle image; acquiring a depth image associated with the first visualangle image, and converting the first planar motion vector into a first stereoscopic motion vector based on depth information associated with the moving target in the depth image; converting the firststereoscopic motion vector in a first visual angle into a second stereoscopic motion vector in a second visual angle according to a visual angle conversion matrix between the first visual angle and the second visual angle; converting the second stereoscopic motion vector into a second planar motion vector according to a projection matrix of the second visual angle; and generating an image including the moving target in the second visual angle based on the second planar motion vector. Through the above scheme, system resources can be saved, and the image processing efficiency can be improved.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a multi-view image processing technology including moving objects. Background technique [0002] Game rendering requires a lot of 3D model modeling, rasterization, texture mapping, rendering light and shadow effects, etc. The rendering complexity is very high, so it often causes frame rate bottlenecks. [0003] The prior art mainly aims at the complete video sequence, calculates the non-key frame motion vector off-line, and realizes the interpolation between frames. Such methods have poor real-time performance and complex algorithms, and have not been optimized for mobile platforms. In addition, the existing technology mainly uses color information to analyze images, and does not fuse depth data to reduce redundant calculations and improve efficiency; at the same time, it does not fuse depth data, and adopts multi-scale image processing methods for different depths. Fi...

AI Technical Summary

Problems solved by technology

Such methods have poor real-time performance and complex algorithms, and have not been optimized for mobile platforms

In addition, the existing technology mainly uses color information to analyze images, and does not fuse depth data to reduce redundant calculations and improve efficiency; at the same time, it does not fuse depth data, and uses multi-scale image processing methods for different depths

Finally, the existing technology mainly focuses on the accuracy and rapidity of optical flow calculation in color space, and does not propose the fusion method of depth data and the application in image interpolation

A large number of repetitive textures in color space and violation of optical flow assumptions will have a great impact on motion prediction

The existing technology is mainly aimed at the offline rendering of complete video data, or the online prediction of monocular video data, and does not involve the fusion of multi-objective and multi-dimensional information

Images