Automated Image Keying via Color Space Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current media editing applications face difficulties in accurately isolating and modifying specific color portions in images or videos, particularly when dealing with blue or green screens, due to challenges in tracing boundaries and identifying accurate color spaces, which can result in halo effects and imperfect edits.
Innovation Solution
A novel keyer system that identifies and selects portions of a color space based on pixel samples, allowing users to define a key and transition region through user interface tools, enabling precise editing by applying edits based on the extent of pixel selection within these regions, with features like auto-fitting and three-dimensional volume propagation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a user traces boundaries of an item in an image to select a portion, then the user can modify that portion, but it is difficult for a user to accurately trace the boundaries
Solution Approach 1:
The patent replaces the manual mechanical tracing operation with an automated color-space-based selection system. Instead of requiring users to manually trace boundaries with a cursor, the system automatically identifies and selects pixels based on their color values matching the sampled key color, thereby eliminating the difficulty of accurate boundary tracing while maintaining selection precision.
Solution Approach 2:
The system enables self-service selection by automatically identifying and selecting pixels that match the key color criteria without requiring user intervention for boundary definition. The automated selection process serves itself by using color space analysis to determine which pixels should be included in the selection, freeing the user from the complex task of manual boundary tracing.
2Measurement precision
If the key is defined as a portion of a color space, then color-based selection is enabled, but identifying an accurate portion of the color space is difficult, especially when accounting for halos
Solution Approach 1:
The patent changes the parameters used to define the key portion in color space by introducing multiple sampling points and using statistical analysis (mean and standard deviation) to determine the key color values. This approach transforms the complex problem of identifying an accurate color space portion into a more manageable process of sampling and statistical calculation, thereby improving accuracy while reducing the perceived complexity for the user.
Solution Approach 2:
The system performs preliminary sampling of multiple pixels to establish the key color values before final selection is made. By pre-sampling and analyzing the color values of multiple pixels to determine the mean and standard deviation, the system prepares the color space definition in advance, which simplifies the subsequent selection process and improves accuracy by accounting for variations such as halos.
3Manufacturing precision
If manual tracing is used to select image portions, then specific areas can be modified, but the process is time-consuming and imprecise
Solution Approach 1:
The patent replaces the time-consuming manual tracing mechanism with an automated color-space-based selection system. The system automatically identifies and selects pixels based on their color values, eliminating the need for users to manually trace boundaries. This substitution dramatically reduces the time required for selection while simultaneously improving precision by using objective color value comparisons rather than subjective manual boundary definition.
Solution Approach 2:
The selection process serves itself by automatically identifying pixels that match the key color criteria without requiring user intervention for each pixel or boundary definition. The system independently performs the selection based on color space analysis, which both reduces the time required and improves precision by consistently applying the color matching criteria to all pixels.
4Ease of operation
If a simple color range selection is used, then ease of operation is improved, but accuracy in handling complex color variations and halos deteriorates
Solution Approach 1:
The patent enhances the simple color range selection by changing the parameters from a single color value to multiple sampling points with statistical analysis (mean and standard deviation). This allows the system to account for color variations and halos while maintaining the simplicity of the user interface. The user still performs a simple sampling action, but the underlying parameter analysis is more sophisticated, thereby improving accuracy without sacrificing ease of operation.
Solution Approach 2:
The system performs preliminary sampling of multiple pixels to establish the key color values before final selection is made. By pre-sampling and analyzing the color values to determine the mean and standard deviation, the system prepares a more accurate color range that accounts for variations such as halos. This preliminary action maintains ease of operation (the user still only needs to sample) while improving accuracy through the statistical analysis.
Data Source
AI summary
Some embodiments provide a method for automatically selecting a portion of an image that includes several pixels, each of which has a set of pixel values. The method identifies a background color of the image. For each region of a set of regions of a color space that correspond to the background color, the method determines whether a threshold number of pixels in the image have pixel values that are in the region. The method identifies the pixels of the image in the regions that correspond to the background color and have a threshold number of pixels in the image. The method generates a portion of the color space to define a selection of the image using the pixel values of the identified pixels.