Weighted Face Region Processing for Smooth Image Transitions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies often result in uneven transitions and obvious color and brightness discontinuities when enhancing face regions in images, particularly when multiple faces overlap, leading to streak-shaped cut lines and discontinuities.
Innovation Solution
An image processing system and method that utilizes an object detection module to detect faces based on a polygon shape and a weight calculation module to assign weights to pixels based on their coordinates and face positions, reducing non-face regions and enabling precise brightness and chromaticity adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If face region enhancement is applied to improve face quality, then face brightness and chromaticity are improved, but uneven transitions and color/brightness discontinuities appear at face edges and overlapping regions
Solution Approach 1:
The patent applies different processing weights to different regions within and around face boundaries. Pixels inside face regions receive full enhancement weight, while pixels in overlapping or non-face regions receive reduced or zero weight. This local differentiation ensures that enhancement is concentrated where needed (on actual face pixels) while avoiding the harmful effects of enhancing non-face regions, thus improving both brightness and edge transition uniformity.
2Illumination intensity
If traditional face enhancement methods are used to improve face appearance, then face color and brightness are enhanced, but obvious streak-shaped cut lines and discontinuities appear when multiple faces overlap
Solution Approach 1:
The patent dynamically adjusts the enhancement weight parameter for each pixel based on its spatial location relative to face boundaries and overlapping regions. By changing the weight parameter from 1.0 (full enhancement) to 0.0 (no enhancement) for pixels in non-face or overlapping regions, the system eliminates color discontinuities and cut lines while maintaining chromaticity enhancement in valid face regions.
3Productivity
If rectangular face region marking is used to simplify processing, then processing speed is improved, but large areas of non-face regions are incorrectly included as face regions
Solution Approach 1:
The patent segments the face region from the bounding box by identifying and excluding non-face pixels within the rectangular region. Instead of treating the entire bounding box as a face region, the system divides it into face pixels (to be enhanced) and non-face pixels (to be excluded or given zero weight). This segmentation maintains processing efficiency while significantly improving face region accuracy.
Data Source
AI summary
An image processing system and an image processing method are provided. The image processing method includes: executing the following steps by an object detection module: receiving an image and detecting the image based on an object detection algorithm; and in response to detecting faces, obtaining a range of each face based on a shape of a polygon; executing the following steps by a weight calculation module on a plurality of pixels in a region covered by the ranges of all faces: based on coordinate information of a current pixel among the pixels and position information of each face, obtaining a weight of the current pixel; and in response to there being an unselected pixel among the pixels, selecting the unselected one of the pixels as the current pixel and returning to the aforementioned step of obtaining a weight of the current pixel.