3D Image Text Depth Correction via Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D display technologies face challenges in accurately calculating disparity for overlaid text, leading to text distortion and flickers due to spatial disparity discrepancies, which are particularly pronounced in text areas, making it difficult for viewers to enjoy 3D images without discomfort.
Innovation Solution
An image processing apparatus and method that includes a depth estimation unit, a text area detection unit, a mask generation unit, and a depth correction unit to correct the depth of text areas in 3D images by generating a text mask and applying a uniform depth value, reducing flickers and distortion through depth temporal smoothing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing stereo matching algorithms are used to calculate disparity for overlaid text, then the overall 3D image processing can be performed, but text distortion and flickers occur due to spatial disparity discrepancy
Solution Approach 1:
The patent segments the 3D image into different regions: text areas and non-text areas. A text area detection unit identifies text regions, and a depth correction unit applies different depth correction methods to text areas compared to non-text areas. This segmentation allows the system to maintain overall 3D processing capability while correcting text-specific disparity issues separately.
Solution Approach 2:
The patent applies local quality by detecting text areas and applying specialized depth correction only to those regions. The depth correction unit calculates depth values differently for text areas (using text line information and color filtering) versus other image regions, ensuring text readability while preserving the 3D effect of the background image.
2Manufacturing precision
If depth correction is applied to text areas, then text distortion is reduced, but processing complexity increases due to additional detection and correction steps
Solution Approach 1:
The patent performs preliminary actions by first detecting text areas and generating text masks before applying depth correction. The text area detection unit identifies text regions and creates corresponding masks in advance, which then guide the depth correction process. This preliminary segmentation simplifies the subsequent correction operations by focusing computation only on identified text regions.
Solution Approach 2:
The patent introduces text masks as intermediary structures that mediate between text area detection and depth correction. The mask generation unit creates binary masks representing text regions, which then serve as guides for the depth correction unit to apply appropriate correction only where needed. This intermediary approach organizes the complex processing into manageable stages.
3Ease of operation
If uniform depth value is applied to text areas, then text readability is improved, but 3D depth variation in text regions is lost
Solution Approach 1:
The patent applies local quality by setting uniform depth values specifically for text areas while maintaining varied depth values for non-text areas. The depth correction unit calculates a uniform depth for all text pixels within a detected text region, ensuring consistent text rendering and readability, while preserving the 3D depth variation of the background image through separate processing.
Solution Approach 2:
The patent segments the image into text and non-text regions, applying different depth processing strategies to each. Text regions receive uniform depth correction for readability, while non-text regions maintain their calculated depth variations. This segmentation allows simultaneous optimization of both text clarity and 3D spatial composition.
Data Source
AI summary
An image processing apparatus and a method of processing an image are provided. The image processing apparatus includes a depth estimation unit for estimating a depth of an input three-dimensional (3D) image; a text area detection unit for detecting a text area included in the 3D image; a mask generation unit for generating a text mask for the text area; and a depth correction unit for correcting a depth of the text area based on the estimated depth of the input 3D image and the text mask.