Image Processing Apparatus Motion Vector Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The 'looping path problem' occurs when joining moving images captured along a free two-dimensional camera trajectory, causing misalignment due to accumulation of errors in motion vectors, which is difficult to address without high-precision sensors, leading to deterioration of textures and increased calculation costs.
Innovation Solution
An image processing apparatus that derives motion vectors between images, corrects cumulative errors by distributing them evenly across frames, and writes textures into a frame memory, eliminating the need for subpixel processing and high-precision sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If motion vectors are used to join moving images along a free two-dimensional camera trajectory, then the ability to create panoramic images is improved, but cumulative errors cause misalignment and deterioration of textures
Solution Approach 1:
The patent performs preliminary detection of subjects in the first and second images to identify their positions before the joining process. This preliminary action allows the system to establish reference points that will be used to correct cumulative errors in motion vectors, thereby maintaining alignment precision while preserving trajectory flexibility.
Solution Approach 2:
The patent implements a feedback mechanism where the detected subject positions from the first and second images are used to calculate correction amounts for cumulative errors. These correction amounts are then applied to adjust the joining positions of intermediate images, creating a closed-loop system that compensates for drift and maintains alignment accuracy throughout the panorama creation process.
2Manufacturing precision
If high-precision sensors are used to correct cumulative errors in motion vectors, then image alignment precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent enables the image processing apparatus to self-correct cumulative errors by using its own image data. The system detects subjects within the images themselves and uses these detected positions as reference information to calculate correction amounts, eliminating the need for external high-precision sensors while maintaining alignment precision through self-service error correction.
Solution Approach 2:
The patent creates a virtual reference system by detecting and copying subject positions from the first and second images. These copied position data serve as reference information that replaces the function of physical high-precision sensors, allowing the system to achieve accurate alignment without adding complex hardware components.
3Manufacturing precision
If subpixel processing is used to reduce misalignment, then image alignment precision is improved, but calculation costs increase
Solution Approach 1:
The patent segments the error correction process by applying corrections in two stages: first correcting drift errors using subject position references, and then separately addressing misalignment errors. This segmentation allows the system to handle different types of errors with appropriate methods, reducing the overall calculation burden compared to applying subpixel processing uniformly to all correction needs.
Solution Approach 2:
The patent applies correction amounts derived from subject position detection to a selected number of pixels around the joining position rather than performing exhaustive subpixel processing on all pixels. This partial action approach achieves sufficient alignment precision for practical purposes while significantly reducing calculation costs compared to complete subpixel processing.
Data Source
AI summary
Provided are an image processing apparatus, an image processing method, a program, and a camera which are capable of generating a joined image in which misalignment is less likely to occur. An image processing apparatus according to an exemplary embodiment includes a motion vector derivation unit that derives a motion vector between images of a plurality of images; a correction unit that corrects a motion vector between images included in a target section from a first image to a second image based on cumulative errors in the section from the first image to the second image; and a texture writing unit that writes, into a frame memory, the plurality of textures that form the joined image, based on the motion vector corrected by the correction unit.