Image Transforming Device Using Fixed Grid Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face transforming functions in digital image capturing devices require increased processing load due to the need to rotate division points or facial contours based on the inclination of the face, leading to longer processing times, especially in small-sized devices like digital cameras and camera mobile phones.
Innovation Solution
An image transforming device that sets control points and grid points based on characteristic information, allowing the grid points to remain stationary regardless of the inclination of the control points, thereby reducing the need for area determination and processing load, using a movement amount vector distribution model or manual setting for control points and interpolating movement amounts for grid points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If division points or facial contour are arranged in accordance with detected facial image and rotated to match face inclination, then image transformation accuracy is improved, but processing load increases
Solution Approach 1:
The patent segments the image transformation process into two independent parts: control points that follow face inclination and grid points that remain fixed. This segmentation allows the grid points to avoid rotation calculations while control points adapt to face orientation, thereby reducing processing load while maintaining transformation accuracy.
Solution Approach 2:
Instead of rotating all grid points to match face inclination as in conventional methods, the patent inverts the approach by keeping grid points fixed and only rotating control points. This inversion eliminates unnecessary rotation calculations for grid points while still achieving accurate image transformation through the control point movement.
2Manufacturing precision
If grid points are rotated to match control point inclination, then image area transformation accuracy is improved, but processing time increases
Solution Approach 1:
The patent divides the point system into control points and grid points with different functions. Control points are rotated to match face inclination while grid points remain fixed, segmenting the rotation operation to eliminate unnecessary calculations and reduce processing time.
Solution Approach 2:
The patent extracts the rotation requirement from the grid points and applies it only to control points. By taking out the rotation operation from the general grid point system and applying it selectively only where needed (at control points), processing time is reduced while maintaining transformation accuracy.
3Manufacturing precision
If boundary determination and area inclusion checks are performed for each pixel, then transformation precision is improved, but processing load increases
Solution Approach 1:
The patent segments the image into quadrangular regions defined by fixed grid points. This segmentation creates a structured framework that simplifies pixel assignment and boundary determination, reducing processing load while maintaining transformation precision through the organized quadrangular division.
Data Source
Figure 1
Figure 2(a)~2(c)
Figure 3~4
AI summary
A control section (11) of a digital camera (10) transforms at least a part of a captured image of a face based on a predetermined type of transformation. The control section (11) includes: an image obtaining section (20) which obtains the captured image of the face; a characteristic point detecting section (22) which detects a characteristic point from the captured image of the face, the characteristic point characterizing the face; a control point setting section (23) which sets control points and a movement amount vector of the control points for the captured image, based on the detected characteristic point and on an arrangement pattern and a movement pattern each corresponding to the predetermined type of transformation; a grid point setting section (24) which sets grid points for the captured image, and sets a movement amount vector of the grid points based on the control points and the movement amount vector of the control points; and an image transforming section (25) which transforms an image are by moving the grid points based on the movement amount vector of the grid points, the image area being a quadrangle, the grid points being four vertices of the quadrangle. According to this, it is possible to reduce a processing load in image transformation.