Importance

Loading Importance Indicators...

Abstract

A system and method automatically detects and corrects the occurrence of red-eye in digital photographs and images without user intervention. The system includes a face recognition and locator engine for locating human faces within the image and for generating a bounding box around each face. A red-eye locator engine analyzes the pixel data for each bounding box and computes one or more predefined metrics. The preferred metrics include color variation, redness, redness variation and glint. The red-eye locator engine also generates one or more detection masks based upon the computed metrics, and searches the detection mask for an occurrence of red-eye. A red-eye correction engine receives the detection mask including the detected occurrences of red-eye, and generates a correction mask. Pixels identified as being occurrences of red-eye are then de-saturated in accordance with the correction mask.

Loading the Abstract Image...

First Claim