Lip Color Correction via Proximity and Similarity Weight Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to accurately correct the color of lips in images due to individual differences in lip color, leading to unnatural or inaccurate lip region detection and correction.
Innovation Solution
An image processing apparatus that generates a lip correction map by calculating proximity and color similarity values for each pixel, integrating these values to determine a distribution of weights for natural color correction, and combining a predetermined correction color with pixel hues using weights defined in the map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If lip region is specified based on lip color alone, then the processing is simple, but the lip region cannot be accurately specified due to individual differences in lip color
Solution Approach 1:
The patent segments the lip region specification process into multiple independent components: (1) initial lip region specification based on color, (2) teeth region specification based on color similarity to teeth, (3) exclusion of teeth region from lip region, and (4) generation of distance maps. This segmentation allows each component to be processed independently and accurately, resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent introduces teeth region detection as an intermediary step to improve lip region accuracy. By detecting the teeth region based on color similarity and excluding it from the lip region, the system achieves more accurate lip boundary specification. This intermediary approach resolves the contradiction by adding a processing step that significantly improves accuracy without excessive complexity.
2Ease of operation
If uniform correction is applied to the entire detected lip region, then the processing is straightforward, but the correction does not account for individual differences in lip shape and color distribution
Solution Approach 1:
The patent applies local quality by generating a distance map that assigns different correction weights to different regions within the lip area. Pixels closer to the lip boundary receive different correction intensity than pixels in the center, allowing the correction to adapt to individual lip shapes and color distributions. This resolves the contradiction by making the correction process adaptive to local characteristics while maintaining a systematic approach.
Solution Approach 2:
The patent introduces dynamic adaptation by calculating distance values from each pixel to the lip boundary and using these distances to determine correction weights. This dynamic approach allows the correction to automatically adapt to different lip shapes, sizes, and color distributions without requiring manual adjustment, resolving the contradiction between straightforward operation and adaptability to individual differences.
3Area of stationary object
If the lip region is expanded to include all reddish areas, then more of the lip is covered, but the region becomes larger than the actual lip due to including non-lip areas
Solution Approach 1:
The patent extracts and removes the teeth region from the initially detected lip region by comparing color similarity to teeth representative color. This extraction process removes non-lip areas (such as teeth with reddish tones) from the lip region, reducing the area to accurately match the actual lip boundaries while maintaining coverage of the entire lip area.
Solution Approach 2:
The patent uses feedback by calculating distance values from each pixel to the lip boundary and using this distance information to refine the correction application. The distance map provides feedback about the spatial relationship between pixels and the lip boundary, allowing the system to adjust correction intensity to achieve accurate lip boundary detection while maintaining appropriate coverage.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing apparatus comprises an image acquiring unit configured to acquire a face image including a face of a person; a lip extracting unit configured to extract a lip region from the face image; a color determining unit configured to determine a representative color of the lip based on pixels included in the lip region; a value acquiring unit configured to acquire a first value representing a proximity to the lip region and a second value correlated to similarity of a color to the representative color; a generating unit configured to generate, based on distributions of the first and second values, a lip correction map representing a distribution of weights used in correcting a color of the lip included in the face image; and an image correcting unit configured to correct the color of the lip using the lip correction map.