Cyberlink Corporation

Perfect Corp.

Hao-Ping Hung

Wei-Hsin Tseng

Ho-Chao Huang

Chih-Chao Ma

Pei-Ying Lu

An electronic device and associated method place a sticker near a facial region in a digital image. The method detects 2D positions of facial features from a 2D digital image, and calculates a projection matrix from a predetermined 3D reference model having predefined facial feature points that correspond to the 2D detected facial features. The method then selects a digital sticker, and for each corner of the selected digital sticker, uses the projection matrix to transform 3D positions of the corner to corresponding positions on the 2D digital image. The method calculates a refinement matrix defining a correlation of each corner of the selected digital sticker to anchor points in the 2D digital image. Using the refinement matrix, the method calculates updated projected 2D positions and displays the selected sticker on the 2D digital image based on the updated projected 2D positions for each corner point.

System and method for displaying graphical effects based on determined facial positions

A method implemented in an image editing device comprises determining a source region in a source image and determining a target region in a target image. At least one image characteristic of each of the source region and a region outside the target region is analyzed. The content in the source region is adjusted according to the at least one image characteristic of the source region and the at least one image characteristic of the region outside the target region. The adjusted content from the source region is inserted into the target region.

Systems and methods for seamless patch matching

Systems and methods for performing multimedia conversion are disclosed. In a multimedia conversion system for converting two-dimensional (2D) multimedia data to stereoscopic multimedia data, an absolute difference between pixel pairs is determined for each scan line in a frame of the 2D multimedia data. The absolute difference values are accumulated, and a depth map is derived based on the accumulated difference values. The depth map is applied to the frame in the 2D multimedia data.

Systems and methods for generating a depth map and converting two-dimensional data to stereoscopic data

Various embodiments are disclosed for performing inpainting. One embodiment is a method for editing a digital image in an image editing device. The method comprises obtaining an inpainting region in the digital image, determining a target resolution for scaling a resolution of the digital image based on an original resolution of the digital image, and determining an intermediate resolution level for scaling a resolution of the digital image based on the target resolution. The method further comprises scaling the resolution of the digital image to the intermediate resolution level, performing partial inpainting of the inpainting region at the intermediate resolution, and performing inpainting on a remainder portion in the inpainting region at a final target resolution.

Systems and methods for multi-resolution inpainting

To modify a facial feature region in a video bitstream, the video bitstream is received and a feature region is extracted from the video bitstream. An audio characteristic, such as frequency, rhythm, or tempo is retrieved from an audio bitstream, and the feature region is modified according to the audio characteristic to generate a modified image. The modified image is outputted.

Method of facial image reproduction and related device

A method of processing interlaced video data including a first interlaced field and a second interlaced field is provided. The method includes performing a deinterlacing operation upon the interlaced video data to generate a first deinterlaced frame corresponding to the first interlaced field and to generate a second deinterlaced frame corresponding to the second interlaced field; performing motion estimation according to video information derived from the interlaced video data to generate a motion estimation result; and performing a blending operation upon the first deinterlaced frame and the second deinterlaced frame to generate an output frame. The blending operation is based on the motion estimation result, and the output frame replaces the second deinterlaced frame. Specifically, the first interlaced field is immediately followed by the second interlaced field.

Growing community of inventors

Taipei, Taiwan

Hao-Ping Hung