Projection Image Correction with Two Feature Points and Surface Normals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for deriving projective transformation parameters in projectors require four or more feature points, which can be difficult to find due to obstacles, view angles, and screen shapes, making accurate projection correction challenging.
Innovation Solution
A method and system that uses an imaging device to capture images before and after a projector's position change, extracting feature points and normal vectors to correct the projection image's position based on these points and vectors, reducing the complexity of matrix calculations by using two feature points per image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If four or more feature points are used to derive projective transformation parameter, then the transformation accuracy is improved, but the difficulty of finding sufficient feature points increases due to obstacles, view angles, and screen shapes
Solution Approach 1:
The patent extracts only the essential elements needed for transformation - specifically two feature points and their corresponding normal vectors - rather than requiring four or more feature points. This extraction of minimal necessary information reduces the difficulty of finding sufficient feature points while maintaining transformation accuracy through the addition of normal vector data.
Solution Approach 2:
The patent transitions from using only two-dimensional feature point coordinates to incorporating three-dimensional normal vector information. By adding this dimensional element (the normal vectors that define the planar surface orientation), the system achieves accurate projective transformation with fewer feature points, as the normal vectors provide additional geometric constraints that compensate for the reduced number of feature points.
2Measurement precision
If four or more feature points are required for projective transformation, then the transformation parameter derivation is more accurate, but the system complexity and computational load increase
Solution Approach 1:
The patent extracts and uses only the minimal necessary data elements - two feature points and two normal vectors - to derive the projective transformation parameter. This streamlined approach reduces system complexity by eliminating the need to manage, store, and process data from four or more feature points, while maintaining transformation accuracy through the inclusion of normal vector information.
3Measurement precision
If four or more feature points are used, then the projective transformation parameter can be derived accurately, but the time required for parameter derivation and projection correction increases
Solution Approach 1:
The patent extracts and processes only two feature points and their corresponding normal vectors, significantly reducing the computational workload compared to processing four or more feature points. This minimal data extraction approach accelerates the parameter derivation process while maintaining accuracy through the geometric constraints provided by the normal vectors.
Solution Approach 2:
The patent uses exactly two feature points - the minimum necessary when combined with normal vector information - rather than the four or more points traditionally required. This partial action approach (using fewer points than conventional methods) reduces computation time while the added normal vector data compensates to maintain transformation accuracy.
Data Source
AI summary
A correction method includes acquiring a first captured image obtained by imaging of a projection image and a planar surface in a first period, acquiring a second captured image obtained by imaging of the projection image and the planar surface in a second period after the first period, extracting a first feature point and a second feature point of the planar surface in the first captured image, extracting a third feature point corresponding to the first feature point and a fourth feature point corresponding to the second feature point of the planar surface in the second captured image, calculating a first normal vector of the planar surface in the first period, calculating a second normal vector of the planar surface in the second period, and correcting a position of the projection image with respect to the planar surface in the second period to a position of the projection image with respect to the planar surface in the first period based on a set of the first feature point, the second feature point, and the first normal vector and a set of the third feature point, the fourth feature point, and the second normal vector.