Omnivision Technologies, Inc.

Wenshou Chen

Yiyi Ren

Guansong Liu

Alireza Bonakdar

Badrinath Padmanabhan

Richard Mann

Lei Fan

Boyd Fowler

Ramakrishna Kakarala

Lu Chang

Chin Poh Pang

Xinting Gao

Shan Xu

Yanfang Huang

Badri Padmanabhan

Nian Xiong

A method for full interpolating green pixels from an input image having a first minimum repeating unit comprising 4×4 pixels, where 2×2 pixels of same color are grouped together, comprises down sampling of the input image to a first down sampled image, down sampling of the input image to a second down sampled image, and interpolating green pixels resulting in an interpolated green down sampled image, where the interpolating uses jointly the first down sampled image and the second down sampled image. The interpolated green pixels in an interpolated green down sampled image are further up sampled resulting in a full interpolated green image.

Full green interpolation in remosaicing Bayer pattern

An imaging system comprising a phase-detection image sensor comprising a plurality of phase-detection pixel units and a processor configured to: interpolate a green image to obtain a full resolution interpolated green image including defocused portions having artifacts and in-focus portions having sharp image, low-pass filter the full resolution interpolated green image to obtain a blurred image of the interpolated green image, combine the full resolution interpolated green image and the blurred image of the full resolution interpolated green image to obtain a corrected full resolution interpolated green image, where the artifacts of the defocused portions of the full resolution interpolated green image are removed, and the sharp image of the in-focus portions of the full resolution interpolated green image is unaffected.

Phase-detection image sensor remosaicing

A method and apparatus for embedding a digital watermark in image content that is not visible to the human eye is performed on single-sensor digital camera images (often called 'raw' images) from a pixel-array. The raw image is transformed to generate preprocessed image coefficients, a watermark message is encrypted using a first key; the encrypted watermark message is randomized using a second key to form a watermark; and the watermark is embedded in randomly selected preprocessed image coefficients.

Sensor watermarking on raw images

In some embodiments, an image sensor is provided. The image sensor comprises a plurality of photodiodes arranged as a photodiode array. The photodiodes of the photodiode array are arranged into a first quadrant, a second quadrant, a third quadrant, and a fourth quadrant. A first polarization filter and a first telecentric lens are aligned with the first quadrant. A second polarization filter and a second telecentric lens are aligned with the second quadrant. A third polarization filter and a third telecentric lens are aligned with the third quadrant. A fourth telecentric lens is aligned with the fourth quadrant.

Techniques for measuring depth and polarization from a single sensor

An image sensor processor implemented method for retaining pixel intensity, comprising: receiving, by the image processor, a numerical value indicative of a corresponding pixel intensity; determining, by the image processor, whether a least significant portion of the received numerical value is equal to a predetermined numerical value; and responsive to determining the least significant portion of the received numerical value is equal to the predetermined numerical value, rounding, by the image processor, the received numerical value of the corresponding pixel intensity to a higher or lower value depending on a bit sequence, and if the least significant portion of the received numerical value is not equal to the predetermined value, rounding the received numerical value to the higher or lower value based on the received numerical value; and binning the rounded value.

System and method for maintaining pixel intensity when rounding

In some embodiments, an image sensor is provided. The image sensor comprises a plurality of photodiodes arranged as a photodiode array. The plurality of photodiodes includes a first set of photodiodes configured as phase detection photodiodes, and a second set of photodiodes configured as polarization detection photodiodes. In some embodiments, a controller is provided. The controller comprises circuitry configured to process signals from a first set of photodiodes of a photodiode array to obtain depth information; process signals from a second set of photodiodes of the photodiode array to obtain polarization information; process the polarization information to obtain an ambiguous set of surface normals; and process the ambiguous set of surface normals using the depth information to obtain a three-dimensional shape image.

Devices and methods for obtaining three-dimensional shape information using polarization and phase detection photodiodes

A system comprising a polarization CMOS image sensor, at least one processor and a non-transitory computer-readable medium having computer-executable instructions stored thereon that, in response to execution by the at least one processor, cause the system to perform actions including receiving, from the polarization CMOS image sensor, polarization information and two-dimensional image information. The polarization information is processed using a machine learning model to generate an output that indicates whether the polarization information represents a valid biometric measurement of a physical feature of a subject.

Techniques for robust anti-spoofing in biometrics using polarization cues for NIR and visible wavelength band

A multiple camera imaging system comprises a first camera image sensor to obtain a first image of a scene and a second camera image sensor to obtain a second image. The system further comprises an image signal processor (ISP) to produce a first roughly-aligned image from the first image and a second roughly-aligned image from the second image. A feature point in the first roughly-aligned image is identified. A center of a search zone is created, which is a combination of the position of the feature point and an associated disparity value from the first and second roughly-aligned images. The search zone is created within the second roughly-aligned image. Candidate feature points are identified within the search zone, which is much smaller than the entire image. A best-matched feature point pair in the first roughly-aligned image and the second roughly-aligned image is then identified.

Growing community of inventors

Santa Clara, CA, United States of America

Wenshou Chen