Digital Image Color Space Channel Blending for 360-Degree Seam Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing solutions for stitching 360-degree field of view images captured by two 180-degree cameras often result in noticeable seams due to color and intensity differences, leading to blurring and unwanted artifacts.
Innovation Solution
An image blending module applies channel gains in a color space, such as YCbCr or RGB, to enhance pixels and remove seam artifacts, using filtered average values to blend images without introducing blurring or degrading texture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-generated harmful factors
If conventional image processing solutions (averaging technique, alpha-blending technique, Poisson blending, pyramid blending) are used to stitch 360-degree images, then the seam between images can be blended, but blurring and unwanted image artifacts are introduced in the resulting image
Solution Approach 1:
The patent segments the image blending process into distinct color space channels (e.g., luminance channel and chrominance channels in YCbCr color space). By processing each channel separately with appropriate blending strategies, the method avoids the blurring that occurs when blending all pixel values uniformly. The luminance channel can be blended more aggressively to remove seams while chrominance channels are blended more conservatively to preserve sharpness and avoid color artifacts.
Solution Approach 2:
The patent applies different blending weights and strategies to different regions and channels of the image. In the overlap region between two stitched images, the blending ratio is adjusted locally based on color and intensity differences. The method also applies different blending approaches to different color channels, allowing local optimization of both seam removal and sharpness preservation in various parts of the image.
2Object-generated harmful factors
If blending techniques are applied to remove seam artifacts, then the appearance of seams is reduced, but the processing complexity and computational intensity increase
Solution Approach 1:
The patent extracts the color and intensity information from the overlap region between stitched images and uses this extracted information to determine blending parameters. By separating the seam removal problem into distinct color space components and using the extracted color differences to guide blending, the method achieves effective seam removal with simpler processing than full-image complex blending algorithms.
Solution Approach 2:
The patent changes the color space representation (e.g., converting to YCbCr color space) to separate luminance and chrominance information, making the blending process more efficient. By adjusting blending parameters based on color differences in the overlap region and applying channel-specific gains, the method reduces computational complexity while maintaining effective seam removal.
3Object-generated harmful factors
If aggressive blending is applied to remove color and intensity differences along the seam, then seam artifacts are reduced, but image texture and sharpness are degraded
Solution Approach 1:
The patent segments the color information into different channels (luminance and chrominance) in the YCbCr color space. This allows aggressive blending to be applied to the luminance channel for seam removal while the chrominance channels are blended more conservatively to preserve color sharpness and texture details, preventing the degradation that occurs with uniform aggressive blending.
Solution Approach 2:
The patent applies partial blending action by using different blending ratios for different channels and regions. Instead of applying uniform aggressive blending across all image data, the method applies moderate blending to chrominance channels and more aggressive blending only where necessary in the luminance channel, thus removing color mismatches while preserving texture quality.
Data Source
AI summary
In aspects of digital image color space channel blending, a camera device can capture digital images that encompass a three-hundred and sixty degree (360°) field of view. An image blending module is implemented to combine the digital images along a seam between the digital images to form a blended image. To combine the digital images, the image blending module can determine mismatched color between the digital images along the seam within an overlap region that overlaps two of the digital images along the seam. The image blending module can then blend the digital images by channel gains in a color space applied to enhance pixels of one of the digital images starting within the overlap region along the seam and blending into the one digital image.