Image Sensor Conditional Read-Out for Low-Light Sensitivity

The patent implements conditional read-out operations where only pixels exceeding a threshold are fully read out, while other pixels undergo partial read-out or are skipped entirely. This partial action approach reduces power consumption by avoiding unnecessary read-out operations on pixels that do not contribute meaningful signal, while still capturing sufficient information for low-light imaging through statistical processing of the subset of read pixels.

The patent segments the pixel array into groups that share read-out circuitry, enabling selective activation of read-out paths based on scene requirements. By organizing pixels into blocks with shared floating diffusion nodes and read-out circuits, the system can activate only the necessary segments for current imaging conditions, reducing overall power consumption while maintaining sensitivity for pixels that require full read-out.

The patent merges adjacent pixels into blocks that share floating diffusion nodes and read-out circuitry. By combining signals from multiple pixels within each block and applying statistical processing, the system recovers spatial resolution information even when not all individual pixels are fully read out. The merging approach allows the system to maintain effective spatial resolution through signal integration while reducing the number of active read-out paths, thereby lowering power consumption.

The patent dynamically changes operational parameters including read-out threshold levels, pixel sampling density, and block activation patterns based on scene characteristics. By adjusting these parameters in real-time, the system optimizes the balance between power consumption and spatial resolution preservation, adapting the degree of pixel sampling to match scene complexity and lighting conditions.

The patent implements partial read-out operations where only a subset of pixels are fully processed in each frame based on threshold criteria. By performing partial read-out on pixels that meet significance thresholds and using statistical reconstruction for other pixels, the system maintains image information completeness for critical regions while significantly reducing overall read-out time, thereby increasing frame rate without substantial loss of image information.

The patent employs periodic full read-out operations interleaved with partial read-out frames. Rather than reading all pixels every frame, the system periodically performs complete read-outs to update reference information while using partial read-outs for intermediate frames. This periodic action pattern maintains image information completeness over time while achieving higher average frame rates through reduced read-out operations during non-periodic frames.