Digital Image Perspective Transform Using Preview Region Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital image processing methods experience temporal delays when performing perspective transformations on images, as they require detecting a rectangular region after capturing an image, which can hinder efficient image manipulation in applications like healthcare.
Innovation Solution
A digital image processing method that detects and determines a main rectangular region from a preview image before capturing, using coordinate information to generate a perspective-transformed image, thereby reducing or eliminating temporal delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If rectangular region detection is performed after image capture, then the processing flow is simple, but temporal delay occurs reducing efficiency
Solution Approach 1:
The patent applies preliminary action by detecting the rectangular region and obtaining coordinate information from the preview image before the actual image capture operation. This advance detection allows the perspective transform parameters to be prepared in advance, eliminating temporal delay when the image is captured and processed immediately afterward.
2Measurement precision
If multiple rectangular regions are detected from preview image, then detection accuracy improves, but processing complexity increases
Solution Approach 1:
The patent applies segmentation by detecting multiple rectangular regions from the preview image and then separating them into effective and non-effective regions through filtering based on coordinate information stability. This allows accurate detection of multiple potential regions while managing processing complexity through systematic filtering and selection of the main rectangular region.
Data Source
AI summary
A digital image processing method and apparatus eliminates or reduces the temporal delay associated a perspective transform being generated for a region of an captured image. The method includes: receiving a preview image; detecting at least one rectangular region from the preview image and obtaining coordinate information corresponding to the at least one rectangular region; determining a main rectangular region from among the at least one rectangular region detected from the preview image; capturing an image; and generating a perspective-transformed image by performing perspective transform on the captured image by using coordinate information corresponding to the main rectangular region.