Arcsoft Corporation Limited

Donghui Wu

Lingxiang Zhou

Yiqing Jin

Yushan Huang

Kaixuan Mao

Min Cheng

Wenhua Zhao

Anding Zhu

Zhenyu Yang

Jie Sun

Tianxiang Yao

Chu-Fei Chang

Xing Fan

Yunfang Zhu

Huiqiong Wang

A method for simulating an image captured at a long exposure time ('simulated image'), includes (1) capturing each of first, second, and third images at a short exposure time, (2) determining a first relative motion between the first and the second images, (3) transforming the first image to remove the first relative motion, (4) determining a second relative motion between the third and the second images, (5) transforming the third image to remove the second relative motion, and (6) combining the first, the second, and the third images to form the simulated image. Relative motions between images are determined by matching blocks at multiple resolutions to determine corresponding points between the images. Transformation to remove relative motion is determined by fitting corresponding points between the images using a minimum square error (MSE) algorithm in a random sample consensus (RANSAC) framework.

Algorithm description on non-motion blur image generation project

A method for enhancing a video includes applying a linear histogram stretch, a gamma correction, and sigma row, column, and temporal filters to the video.

Video auto enhancing algorithm

A method for removing a red eye from an image includes (1) calculating a weighted red value for each pixel in the image from red, green, and blue color values and a luminance value of each pixel in the image, (2) selecting a plurality of pixels in the image having weighted red values greater than a threshold as red eye pixels, and (3) correcting some of the red eye pixels to remove the red eye from the image. The weighted red value for a pixel is calculated as follows: wherein f is the weighted red value, r is the red color value, g is the green color value, b is the blue color value, cis a first weight given to the red color value, cis a second weigh given to the green color value, cis a third weight given to the blue color value, and Y is the luminance.

Automatic red eye removal

A method is provided for processing minimum coded units (MCUs) in a Joint Photographic Expert Group (JPEG) bit stream representing an image. A data structure is generated for storing MCU block numbers, corresponding MCU block coordinates, and corresponding MCU block editing lists. When an editing operation to the image is received, the editing operation is saved to a group of the MCU block editing lists of a group of the MCUs affected by the editing operation. Later the MCUs are retrieved from the JPEG bit stream one at a time and then processed according to the editing operations in the corresponding MCU blocking editing lists.

JPEG processing engine for low profile systems

A method for enhancing an image includes applying an auto-level transformation to an original image to form a first image, applying an auto-contrast transformation to the original image to form a second image, and applying an auto-brightness transformation to the original image to form a third image. The method further includes applying a color cast correction to the first, the second, and the third images to generate a first group of images, applying a gamma correction to the first group of images to generate a second group of images, applying a sharpening correction to the second group of images to generate a third group of images, and presenting the third group of images to a user to select a final result.

Better picture for inexpensive cameras

A method for correcting a radial distortion in an image includes receiving a radial distortion parameter from the user and calculating new intensity values of points in the image to correct a radial exposure distortion in the image.

A method for generating a panoramic image includes receiving a first image, dividing the first image into a first portion and a second portion, rotating the first portion of the first image, saving the rotated first portion of the first image in a nonvolatile memory, receiving a second image, dividing the second image into a third portion and a fourth portion, matching an overlapping region between the second portion of the first image and the third portion of the second image, stitching the second portion of the first image and the third portion of the second image to form a first stitched image, rotating the first stitched image, and saving the first stitched image in the nonvolatile memory.

Panoramic maker engine for a low profile system

A method for removing red eyes includes (1) identifying potential red eye pixels in a low resolution image (hereafter 'LR red eye pixels'), (2) grouping contiguous LR red eye pixels into red eye regions (herafter 'LR red eye regions'), (3) determining working areas around the LR red eye regions (hereafter &#x201c;LR working areas&#x201d;), (4) determining classifiers from the LR working areas, (5) retrieving working areas from a high resolution image that correspond to the LR working areas (hereafter &#x201c;HR working areas&#x201d;), (6) applying the classifiers to the HR working areas to identify potential red eye pixels (hereafter &#x201c;HR red eye pixels&#x201d;), (7) grouping contiguous HR red eye pixels into red eye regions (hereafter &#x201c;HR red eye regions&#x201d;), (8) determining some of the HR red eye regions as the red eyes, and (9) replacing the HR red eye pixels in the red eyes with black pixels.

Enhanced automatic red eye removal

A method for indexing minimum coded units (MCUs) in a Joint Photographic Expert Group (JPEG) bit stream includes (1) entropy decoding a first MCU to determine a bit offset of a second MCU and a DC coefficient of the first MCU, wherein the first MCU immediately precedes the second MCU in the bit stream, and (2) indexing the second MCU by storing the bit offset and the DC coefficient in an index. The method may further include (3) receiving a request for the second MCU, (4) reading the index to determine the bit offset of the second MCU and the DC coefficient of the first MCU, and (5) entropy decoding the second MCU starting at its bit offset in the bit stream, wherein the entropy decoding the second MCU includes determining a DC coefficient of the second MCU using the DC coefficient of the first MCU.

A method for aligning images includes (1) selecting a pair of overlapping pixels when two second image are aligned with a camera motion; (2) if both pixels are edge pixels, incrementing a correlation value between the two images for the camera motion; (3) if only one of pixels is an edge pixel, decrementing the correlation value between the two images for the camera motion; (4) repeating steps (1) to (3) for other pairs of overlapping pixels in the two images to determine the correlation value for the camera motion. The method further includes (5) repeating steps (1) to (4) for other camera motions; (6) selecting camera motions having large correlation values; (7) determining matched edge counts for members of the group; (8) selecting the camera motion with the largest matched edge count; and (8) aligning the two images with the selected camera motion.

Edge based alignment algorithm

A method for improving image resolution includes (a) selecting a low-resolution pixel in a low-resolution image; (b) generating a vector that characterizes a low-resolution patch about the low-resolution pixel; (c) classifying the low-resolution patch into one of a plurality of possible patterns; (d) if said one pattern is one of a plurality of top ranking patterns having filters, applying a filter of said one of the top ranking patterns to the low-resolution patch to generate high-resolution pixels that correspond to the low-resolution pixel; (e) if said one pattern is not one of the plurality of top ranking patterns, applying an interpolation process to the low-resolution patch to generate the high-resolution pixels; (f) repeating steps (a) to (e) for additional low-resolution pixels to generate additional high-resolution pixels to form a high-resolution image.

Pattern classification and filter design for increasing image resolution

A method for correcting a nonlinear distortion in an image includes receiving an image of a calibration template comprising calibration patterns, detecting at least three points from the calibration patterns, calculating a distortion parameter from coordinates of the at least three points that should align along a straight line, and correcting the nonlinear distortion in the image using the distortion parameter. A method for correcting a radial distortion in an image includes receiving a radial distortion parameter from the user and calculating new intensity values of points in the image to correct a radial exposure distortion in the image.

Systems and methods for generating three-dimensional models of an object use images having unmeasured camera parameters. Camera calibration determines the perspective of the camera from the content of the images. A background having a pattern with a known marks in each image can facilitate determination of the camera parameters. One background pattern includes separated marks having rectangular sections where corners of the rectangular sections provide calibrations points for the camera parameters. The camera parameters can also be determined by matching features of the object in different images and determining differences in perspective from differences in the appearance of the matched features in different images. A combination of projective and metric reconstructions provides robust reconstruction.

3-D reconstruction engine

A method for generating texture includes (1) selecting a target patch to be filled in a image, (2) selecting a sample patch as a candidate for filling the target patch, (3) determining a first difference between a first area surrounding the target patch and a corresponding first area surrounding the sample patch, and a second difference between a second area surrounding the target patch and a corresponding second area surrounding the sample patch, (4) multiplying a larger of the first difference and the second difference with a first weight factor, and a smaller of the first difference and the second difference with a second weight factor, and (5) summing the weighted first difference and the weighted second difference as a distance between the target patch and the sample patch.

Growing community of inventors

Fremont, CA, United States of America

Donghui Wu