Image Processing Apparatus Moving Object Locus Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to clearly display the locus of a moving object from multiple frame images without being affected by background conditions or object characteristics such as number, size, transformation, and color, and face challenges with increased computational costs and accuracy when dealing with multiple moving objects.
Innovation Solution
An image processing apparatus that generates a background pixel value based on a predetermined distribution of input image pixel values, calculates a degree-of-difference between input image pixel values and the background pixel value, and uses this degree-of-difference to generate output image pixel values, allowing for clear display of moving object loci without background interference and adaptable to changes in object characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If simple synthesis of multiple frame images is performed by overlapping, then the synthesis process is simple, but the moving object locus becomes light-colored and unclear
Solution Approach 1:
The patent applies local quality by differentiating between background pixels and moving object pixels through degree-of-difference calculation. Background pixels receive uniform weighting while moving object pixels receive enhanced weighting based on their difference from the background, creating localized quality enhancement where it is most needed for locus clarity.
Solution Approach 2:
The patent changes the weighting parameter dynamically based on the degree-of-difference between current frame pixels and background pixels. Pixels with higher degree-of-difference (moving objects) receive higher weighting coefficients, while background pixels receive lower weighting, thereby enhancing locus visibility without sacrificing synthesis simplicity.
2Measurement precision
If weighting is applied to heavily weight moving object pixels away from mean luminance value, then the moving object locus becomes clearer, but the background distinction becomes difficult under certain image conditions
Solution Approach 1:
The patent employs dynamics by iteratively updating the background image and recalculating degrees-of-difference in multiple passes. This dynamic adjustment allows the system to adapt to varying image conditions, improving both locus clarity and background distinction accuracy by refining the background model based on accumulated frame information.
Solution Approach 2:
The patent implements feedback through the iterative process where the synthesized output is used to update the background image, which then serves as the reference for calculating degrees-of-difference in subsequent iterations. This feedback loop continuously refines the distinction between background and moving objects, enhancing adaptability to different image conditions.
3Measurement precision
If equal weighting is applied to multiple synthesis except in background, then the moving object locus can be displayed clearly, but the rate of single frame contribution decreases as the number of frame images increases
Solution Approach 1:
The patent changes the weighting parameter based on the degree-of-difference metric, allowing moving object pixels to maintain high contribution rates even as the number of frames increases. By dynamically adjusting weights rather than applying equal weighting, the system preserves the visibility and contribution of moving objects across multiple frames.
4Adaptability or versatility
If tracing of moving object is performed with interpolation on tracing failure, then the locus can be displayed in arbitrary fashion, but the method becomes inadaptable to changes in object characteristics and computationally expensive for multiple moving objects
Solution Approach 1:
The patent extracts the background component separately through degree-of-difference calculation, isolating moving objects without requiring complex tracing algorithms. This extraction approach eliminates the need for interpolation and tracing operations, significantly reducing computational cost while maintaining adaptability to various object characteristics and motions.
Data Source
AI summary
An image processing apparatus comprising: background generation means for generating, on each of corresponding pixel positions of more than one input image, a pixel value involved in a predetermined distribution among pixel values of the input images as a background pixel value in the pixel position; degree-of-difference generation means for generating, on each of the pixel positions, a degree-of-difference indicating a degree to which each input image pixel value has a difference from the background pixel value; and output image generation means for generating, on each of the pixel positions, an output image pixel value by reflecting the input image pixel values depending on the degree-of-difference.