Signal processor for adjusting image quality of an input picture signal

What is this patent about?

’950 is related to the field of digital signal processing for imaging devices, specifically addressing the challenges of automatic control systems in video and still cameras. In typical imaging environments, high-intensity light sources or sudden feature changes can skew the statistical data used for automatic exposure, auto-focus, and white balance. Traditional systems often struggle when these disruptive elements appear, either by failing to find an optimal measurement window or by requiring iterative, power-intensive calculations to relocate the detection frame.

The underlying idea behind ’950 is to dynamically sanitize the data used for camera control by identifying and excluding disruptive pixels on an individual basis rather than simply shifting a fixed rectangular frame. Instead of searching for a 'clean' area of the image, the system detects specific coordinates where features—such as luminance or contrast—deviate sharply from their surroundings. By treating these outliers as holes in the measurement mask, the processor can calculate control values from the remaining valid pixels, ensuring the camera logic responds to the actual scene rather than a localized spotlight.

The claims of ’950 focus on a signal processor architecture that integrates a feature detection unit with a specialized region creation logic. The system is designed to scan an input picture signal to pinpoint exactly where pixel features change, subsequently generating a modified extraction region that explicitly removes those coordinates. The independent claim covers the closed-loop control mechanism where a processing unit adjusts the image signal based on evaluated values derived exclusively from this filtered, non-disruptive region.

In practice, the invention works by applying a differential analysis to the signal intensity of target pixels relative to their neighbors. When the difference exceeds a predefined threshold—indicating a sharp edge or a high-luminance point source—the system flags these coordinates. A hold unit stores this position information, allowing the region creation unit to 'carve out' these pixels from the integration zone. This allows the camera to maintain a consistent exposure or focus level for the background and primary subjects, even if a bright light enters the frame.

This approach differs from prior solutions that relied on shifting a rigid ranging frame or selecting from a set of pre-defined sub-windows. By utilizing pixel-level exclusion, the invention avoids the 'trial and error' process of moving a detection window to find a spot without high luminance. Furthermore, the system can be configured to adjust the size of the remaining extraction area to keep the total number of sampled pixels constant, which simplifies the downstream control logic and ensures stable, flicker-free adjustments in real-time video applications.