Techniques for lossless compression of a digital image using prior image context.

Techniques for lossless compression of a digital image using prior image context.

Methods and systems are provided for implementing preprocessing operations and augmentation operations upon image datasets transformed to frequency domain representations, including decoding images of an image dataset to generate a frequency domain representation of the image dataset; performing a resizing operation based on resizing factors on the image dataset in a frequency domain representation; performing a reshaping operation based on reshaping factors on the image dataset in a frequency domain representation; and performing a cropping operation on the image dataset in a frequency domain representation. The methods and systems may further include performing an augmentation operation on the image dataset in a frequency domain representation. Methods and systems of the present disclosure may free learning models from computational overhead caused by transforming image datasets into frequency domain representations. Furthermore, computational overhead caused by inverse transformation operations is also alleviated.

Methods and systems are provided for implementing preprocessing operations and augmentation operations upon image datasets transformed to frequency domain representations, including decoding images of an image dataset to generate a frequency domain representation of the image dataset; performing a resizing operation based on resizing factors on the image dataset in a frequency domain representation; performing a reshaping operation based on reshaping factors on the image dataset in a frequency domain representation; and performing a cropping operation on the image dataset in a frequency domain representation. The methods and systems may further include performing an augmentation operation on the image dataset in a frequency domain representation. Methods and systems of the present disclosure may free learning models from computational overhead caused by transforming image datasets into frequency domain representations. Furthermore, computational overhead caused by inverse transformation operations is also alleviated.

A method of encoding video data values, the method including selectively encoding, via circuitry, a high bit depth control flag and, when the high bit depth control flag is set to indicate high bit depth operation, selectively encoding an extended precision flag to indicate at least extended precision operation of a spatial frequency transform stage and encoding the video data values according to a mode of operation defined by the encoded high bit depth control flag and, when encoded, the extended precision flag.

A method of encoding video data values, the method including selectively encoding, via circuitry, a high bit depth control flag and, when the high bit depth control flag is set to indicate high bit depth operation, selectively encoding an extended precision flag to indicate at least extended precision operation of a spatial frequency transform stage and encoding the video data values according to a mode of operation defined by the encoded high bit depth control flag and, when encoded, the extended precision flag.

A feature data encoding method, an encoder, a feature data decoding method, and a decoder are provided. The feature data encoding method includes following steps. A transform unit is divided into a plurality of sub-blocks and N sub-transform units. A reference origin and a LSC are determined in an i-th sub-transform unit of the sub-transform units, and an original coordinate of the last significant coefficient of the i-th sub-transform unit is modified to a specific coordinate. The i-th sub-transform unit is scanned from a specific sub-block of the i-th sub-transform unit, and significant feature coefficients in the i-th sub-transform unit are individually encoded as coded data.

A feature data encoding method, an encoder, a feature data decoding method, and a decoder are provided. The feature data encoding method includes following steps. A transform unit is divided into a plurality of sub-blocks and N sub-transform units. A reference origin and a LSC are determined in an i-th sub-transform unit of the sub-transform units, and an original coordinate of the last significant coefficient of the i-th sub-transform unit is modified to a specific coordinate. The i-th sub-transform unit is scanned from a specific sub-block of the i-th sub-transform unit, and significant feature coefficients in the i-th sub-transform unit are individually encoded as coded data.

Automotive radar systems and methods include a radar monolithic microwave integrated circuit (MMIC) configured to perform radar processor functionality including performing range fast Fourier transforms (FFTs) on a plurality of received radar signal streams to obtain a plurality of transformed radar signal streams, performing H264/H265 encoding on I-frames of the plurality of transformed radar signal streams to obtain a plurality of compressed radar signal streams, and outputting, via a network interface, the plurality of compressed radar signal streams, and a domain controller connected to the radar MMIC via the network interface and configured to receive and utilize the plurality of compressed radar signal streams for an advanced driver-assistance system (ADAS) or autonomous vehicle driving feature, wherein the automotive radar systems/method do not include or utilize a distinct or standalone radar processor.

A method for encoding a picture of a video sequence in a bit stream that constrains slice header processing overhead is provided. The method includes computing a maximum slice rate for the video sequence, computing a maximum number of slices for the picture based on the maximum slice rate, and encoding the picture wherein a number of slices used to encode the picture is enforced to be no more than the maximum number of slices.