Abstract: An imaging device includes an image sensor and an image processor. The image sensor includes an array of pixels. At least one pixel in the array of pixels includes an array of phase disparity subpixels. The array of pixels includes a set of filter elements arranged in a filter pattern, with different subsets of pixels receiving light through different types of filter elements in the set of filter elements. The image processor is configured to receive values measured by at least the phase disparity subpixels; estimate phase disparities between the received values; generate, using the estimated phase disparities, a set of corrected values for the phase disparity subpixels; and generate, using the set of corrected values for the phase disparity subpixels, a set of demosaiced values for each image plane in a set of image planes. Each image plane is associated with a different type of filter element in the set of filter elements.

Abstract: A pulse-width modulation (PWM) image sensor is described herein. The PWM image sensor may have a stacked configuration. A top wafer of the PWM image sensor may have a charge-to-time converter and a logic wafer, stacked with the top wafer, may include a time-to-digital converter. The PWM image sensor may utilize variable transfer functions to avoid highlight compression and may utilize non-linear time quantization. A threshold voltage, as input to a charge-to-time converter, may additionally be controlled to affect light detection, dynamic range, and other features associated with the PWM image sensor.

Abstract: Various implementations disclosed herein include devices, systems, and methods that buffer events in device memory during synchronous readout of a plurality of frames by a sensor. Various implementations disclosed herein include devices, systems, and methods that disable a sensor communication link until the buffered events are sufficient for transmission by the sensor. In some implementations, the sensor using a synchronous readout may select a readout mode for one or more frames based on how many of the pixels are detecting events. In some implementations, a first mode that reads out only data for a low percentage of pixels that have events uses the device memory and a second mode bypasses the device memory based on accumulation criteria such as high percentage of pixels detecting events. In the second mode, less data per pixel may be readout.

Abstract: Various implementations disclosed herein include devices, systems, and methods that determine a control signal before synchronous readout of a frame (or part of the frame) is performed by a sensor. Various implementations disclosed herein include devices, systems, and methods that determine event density (e.g., a dense row signal or a dense frame signal) before synchronous readout is performed by the sensor. In some implementations, the event camera is capable of operating in a first readout mode and a second readout mode based on a number of events. In the second readout mode, less data per pixel may be readout.

Abstract: A method of image data interpolation. The method includes obtaining transform data associated with a plurality of sampling points, each sampling point located on a boundary of a region. The transform data is processed to generate additional transform data associated with an additional sampling point located on a boundary of a subregion of the region. An interpolation process is performed. The interpolation process includes processing image data associated with a point within the subregion with the additional transform data, thereby generating interpolated image data representative of an interpolated data value at the point. This patent application further relates to a method including obtaining transform data associated with points corresponding to a surface of a region and generating additional transform data associated with at least one interior point of the region. Image data associated with a further interior point of the region is interpolated using at least the additional transform data.

Abstract: A method of image data interpolation. The method includes obtaining transform data associated with a plurality of sampling points, each sampling point located on a boundary of a region. The transform data is processed to generate additional transform data associated with an additional sampling point located on a boundary of a subregion of the region. An interpolation process is performed. The interpolation process includes processing image data associated with a point within the subregion with the additional transform data, thereby generating interpolated image data representative of an interpolated data value at the point. This patent application further relates to a method including obtaining transform data associated with points corresponding to a surface of a region and generating additional transform data associated with at least one interior point of the region. Image data associated with a further interior point of the region is interpolated using at least the additional transform data.