Adaptive Depth Offset Calculation for 3D Image Edge Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting depth maps, particularly for 3D displays, fail to accurately capture the overall depth offset of an image, leading to inadequate perspective representation and viewer discomfort due to magnification effects and depth jumps during zooming or scene movement.
Innovation Solution
An adaptive depth offset calculation system that analyzes the edge map of an image by counting high and low frequency points, classifying the image based on predetermined thresholds, and calculating a depth offset using specific equations for different classifications, with optional smoothing techniques to ensure smooth transitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If relative depth map extraction methods are used, then depth information can be obtained based on object saliencies and relative positions, but the overall perspective characteristics and true depth offset of the image cannot be accurately captured
Solution Approach 1:
The patent segments the depth offset extraction process into distinct frequency-based classifications. By analyzing edge maps and separating high-frequency and low-frequency components, the method extracts depth offsets for different image regions independently, thereby improving overall depth accuracy without overwhelming computational complexity
Solution Approach 2:
The patent changes the parameter space by introducing frequency-domain analysis parameters (high-frequency threshold, low-frequency threshold) to classify image regions. This transformation from spatial domain to frequency domain enables more accurate depth offset extraction while maintaining manageable computational requirements through parameter-based classification
2Reliability
If depth offset is not properly adjusted, then image processing is simplified, but perspective perception becomes unrealistic and viewer comfort decreases
Solution Approach 1:
The patent performs preliminary classification of image regions into high-frequency and low-frequency categories before depth offset adjustment. This preliminary action enables the system to apply appropriate depth offset strategies for each region type, ensuring realistic perspective perception while automating the complex adjustment process
Solution Approach 2:
The patent applies different depth offset adjustment strategies to different regions based on their frequency characteristics. High-frequency regions receive one type of adjustment while low-frequency regions receive another, ensuring that each region's depth characteristics are optimized for realistic perspective perception without requiring manual intervention
3Productivity
If fixed depth offset methods are used, then processing is straightforward, but smooth depth transitions in image sequences cannot be achieved
Solution Approach 1:
The patent introduces dynamic adaptation by classifying image sequences into different types (first type and second type) based on their characteristics. The system dynamically selects different depth offset calculation methods for different sequence types, enabling smooth depth transitions while maintaining processing efficiency through automated classification
Data Source
Figure 1
AI summary
Present invention comprises a system and method for adaptive depth offset calculation for an image (F), in which the edge map of the image (F) is analyzed in order to count a number of high frequency points (NHIGH) values, which are higher than a predetermined threshold (EHIGH); and count a number of low frequency points (NLOW) values of which are lower than a predetermined threshold (ELOW). The number of high and low frequency points (EHIGH, ELOW) are compared to predetermined boundaries (THIGH, TLOW) to classify the image (F), and a depth offset is calculated in accordance with said classification. The invention provides a precise depth adjustment for an image (F) in order to obtain a true perspective perception.