Guangdong Oppo Mobile Telecommunications Corp., Ltd.

Wei Quan

Jun Luo

Yuanjiao Ma

An image registration method includes: obtaining a first image and a second image; dividing the first image into blocks to obtain image blocks of the first image; determining, based on a pixel value of each image block, a similar image block of the image block from the second image by invoking a predetermined image block similarity algorithm; calculating, based on position information of the image block and position information of the similar image block, a movement amount of the image block of the first image; performing a clustering on the first image based on a depth value of the first image, the movement amounts of the image blocks, and position information of the first image, to generate clustered image blocks; and moving, when the clustered image blocks satisfy a predetermined condition, the image blocks based on the movement amounts of the image blocks, to obtain a registered image.

Image registration method, terminal, and computer storage medium

An image processing method and apparatus, and a non-transitory computer-readable storage medium are provided. The method includes: acquiring an exposure duration and at least one motion component corresponding to at least one direction within the exposure duration of a current video frame; performing a first noise reduction operation on the current video frame in a two-dimensional space domain in response to a first motion component of the at least one motion component being greater than a preset motion component threshold; and performing a second noise reduction operation on the current video frame in a three-dimensional space domain in response to the at least one motion component being less than the preset motion component threshold.

Image processing method and apparatus, and non-transitory computer readable storage medium

Disclosed is an image processing method, electronic device and storage medium. The method includes obtaining feature information of first region in a current image frame, wherein first region includes a region that is determined by performing motion estimation on the current and previous image frames based on optical flow; obtaining feature information of second region in the current image frame, wherein second region includes a region corresponding to pixel points among first pixel points of the current image frame, where its association with pixel points among second pixel points of the previous image frame satisfies a condition; and based on the feature information of first region and that of second region, fusing the previous and current image frames to obtain a processed current image frame, which is used as a previous image frame for a next image frame.

Growing community of inventors

Yokohama, Japan

Wei Quan