Apple Inc.

Marvellworld Trade Ltd.

Marvell International Limited

Realtek Semiconductor Inc.

Synaptics Corporation

Yun Gong

Jim Chen Chou

Chenge Li

Haiyan He

Dam Le Quang

Guy Cote

I-Ming Pao

Honkai Tam

Yung-Chin Chen

Zhiying Guo

Dam LeQuang

Methods and systems include neural network-based image processing and blending circuitry to blend an output of the neural network to compensate for potential artifacts from the neural network-based image processing. The neural network(s) apply image processing to image data using one or more neural networks as processed data. Enhance circuitry enhances the image data in a scaling circuitry to generate enhanced data. Blending circuitry receives the processed image data and the enhanced data along with an image plane of the processed data. The blending circuitry also determines whether the image processing using the one or more neural networks has applied a change to the image data greater than a threshold amount. The blending circuitry then, based at least in part in response to the change being greater than the threshold amount and/or edge information of the image data, blends the processed data with the enhanced data.

Neural network-based image processing with artifact compensation

A device may include an electronic display to display an image frame based on blended image data and image processing circuitry to generate the blended image data by combining first image data and second image data via a blend operation. The blend operation may include receiving graphics alpha data indicative of a transparency factor to be applied to the first image data to generate a first layer of the blend operation. The blend operation may also include overlaying the first layer onto a second layer that is based on the second image data. Overlaying the first layer onto the second layer may include adding first pixels values of the first image data that include negative pixel values and are augmented by the transparency factor to second pixel values of the second image data to generate blended pixel values of the blended image data.

Image blending modes systems and methods

An electronic device may include scaling circuitry to scale input pixel data to a greater resolution. The directional scaling circuitry may include first interpolation circuitry to receive best mode data, including one or more angles corresponding to content of the image and interpolate first pixel values at first pixel positions diagonally offset from input pixel positions of the input pixel data based on the best mode data and input pixel values corresponding to the input pixel positions. The directional scaling circuitry may also include second interpolation circuitry to receive the best mode data and the input pixel values and interpolate second pixel values at second pixel positions horizontally or vertically offset from the input pixel positions based at least in part on the best mode data and the input pixel values.

Directional scaling systems and methods

Embodiments relate to a super-resolution engine that converts a lower resolution input image into a higher resolution output image. The super-resolution engine includes a directional scaler, an enhancement processor, a feature detection processor, a blending logic circuit, and a neural network. The directional scaler generates directionally scaled image data by upscaling the input image. The enhancement processor generates enhanced image data by applying an example-based enhancement, a peaking filter, or some other type of non-neural network image processing scheme to the directionally scaled image data. The feature detection processor determines features indicating properties of portions of the directionally scaled image data. The neural network generates residual values defining differences between a target result of the super-resolution enhancement and the directionally scaled image data. The blending logic circuit blends the enhanced image data with the residual values according to the features.

Blended neural network for super-resolution image processing

An electronic device may include enhancement circuitry to enhance high resolution image data to improve perceived quality of an image corresponding to the high resolution image data. The enhancement circuitry may include tone detection circuitry to determine one or more tones within the image and apply changes to the high resolution image data based on the one or more tones. The enhancement circuitry may also include example-based improvement circuitry to compare the high resolution image data to low resolution image data and apply changes to the high resolution image data based on differences between sections of the high resolution image data and sections of the low resolution image data. The enhancement circuitry may also include channel processing circuitry to apply the first and second changes to one or more channels of the high resolution image data.

Chrominance and luminance enhancing systems and methods

An electronic device may include angle detection circuitry to receive input image data including a channel of pixel data and determine pixel statistics based on a difference between a first pixel cluster and a second pixel cluster, and the second pixel cluster is offset from the first pixel cluster at a first angle. The pixel statistics may also be based on a difference between the first pixel cluster and a third pixel cluster, wherein the third pixel cluster is offset from the first pixel cluster a second angle, different from the first angle. The angle detection circuitry may also determine a best angle based on the differences, wherein the best angle approximates an angle of uniformity in content of the image. Image processing circuitry may then modify the input image data based on the best angle by modifying pixel values of the channel of pixel data.

Angular detection using sum of absolute difference statistics systems and methods

Systems and methods for generating a detail-enhanced video signal are provided. In a method for generating the detail-enhanced video signal, an input video signal is received. A first signal that is a linear transformation of the input video signal is generated. A detail signal is generated by determining a difference between the input video signal and the first signal, where the detail signal includes information that is added to the input video signal to generate the detail-enhanced video signal. A filtered detail signal is generated by removing noise from the detail signal. The filtered detail signal is multiplied by a gain function to generate a second signal. The gain function is based on an amount of variance in the input video signal. The input video signal and the second signal are combined to generate the detail-enhanced video signal.

Guided filter-based detail enhancement

Systems, methods, and other embodiments associated with efficient visually lossless compression are described. According to one embodiment, an apparatus includes a transform logic configured to receive image data divided into segments and generate a matrix of transform coefficients corresponding to a segment. The apparatus also includes a rate control logic configured to select (i) a set of quantization levels from a plurality of sets of quantization levels and (ii) a quantization level from the selected set of quantization levels based, at least in part, on a desired visual quality. Each set of quantization levels corresponds to a different target compression ratio for the image data. The apparatus further includes a quantization logic configured to quantize the matrix of transform coefficients according to the selected quantization level to produce an array of integers. The apparatus also includes an encoding logic configured to encode the array of integers as a bit stream.

Efficient visually lossless compression

Systems and methods are provided for upscaling a digital image. A digital image to be upscaled is accessed, where the digital image comprises a plurality of pixel values. A first half pixel value is computed for a first point in the digital image based on a plurality of pixel values of the digital image surrounding the first point and an activity level. A second half pixel value is computed for a second point in the digital image, and an interpolated pixel of an upscaled version of the digital image is determined using a plurality of the pixel values, the first half pixel value, and the second half pixel value.

Growing community of inventors

Cupertino, CA, United States of America

Yun Gong