Image Compositing Using Luminance Masks to Reduce Characteristic Color Effects

The patent divides the image into multiple regions using generated masks, separating the target region (e.g., face) from surrounding regions containing characteristic colors like bangs and shadows. This segmentation allows selective processing where the target region is extracted and composited while the harmful characteristic color regions are excluded or treated differently, thus resolving the contradiction between smooth compositing and characteristic color interference.

The patent applies different processing qualities to different regions of the image. The target region receives high-quality processing with preserved detail and color, while surrounding regions containing characteristic colors are masked out or processed with different parameters. This local differentiation ensures that the composite image quality is high where needed while eliminating the harmful effects of characteristic colors from adjacent regions.

The patent performs preliminary processing steps including image normalization, rotation correction, and glass detection before the main face detection process. By preparing the image in advance (straightening rotated faces, removing glass artifacts, standardizing lighting), the subsequent face detection can achieve high accuracy across varied face conditions without requiring complex real-time adaptation during detection.

The patent introduces intermediary processing steps and auxiliary algorithms between the input image and the main detection process. These intermediaries include preprocessing filters, normalization layers, and auxiliary detection models that handle specific challenging cases (glasses, rotation) before the primary face detection algorithm operates, thereby improving overall accuracy and adaptability.

The patent combines multiple processing functions into integrated modules. The mask generation process is merged with luminance adjustment and region specification in a unified processing pipeline, where a single set of processing steps simultaneously achieves multiple objectives (creating masks, adjusting luminance, defining regions) rather than requiring separate independent processing stages, thus reducing overall complexity while maintaining precision.

The patent designs processing modules that perform multiple functions simultaneously. The mask generation algorithm also serves as a region specification tool and luminance adjustment reference, making it a multi-functional component that reduces the need for separate dedicated processing steps for each function, thereby improving region specification accuracy without proportionally increasing processing pipeline complexity.