Warped Object Region Recognition for Low-Angle Floor Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic apparatuses face challenges in maintaining consistent object recognition accuracy due to varying camera angles and distances, particularly struggling with flat objects at long distances or low camera positions, leading to low recognition rates.
Innovation Solution
An electronic apparatus equipped with a sensor, camera, and processor that detects objects, identifies whether they are floor objects, warps object regions based on distance information, and uses AI models trained for specific distances to enhance recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a camera captures objects at long distance or low angle, then the coverage area increases, but the object recognition accuracy decreases
Solution Approach 1:
The patent segments the image processing into multiple stages: initial object detection, distance calculation, warping determination, and final recognition. By dividing the processing pipeline and applying warping only to detected object regions rather than the entire image, the system maintains high recognition accuracy for distant objects while preserving broad coverage capability
Solution Approach 2:
The patent dynamically changes image parameters through warping transformation based on calculated distance and angle parameters. The warping process adjusts the object region's geometric parameters (scale, orientation) to compensate for perspective distortion, enabling accurate recognition regardless of capture distance or angle
2Adaptability or versatility
If the camera position is lowered to capture floor objects, then the ability to detect floor-level objects improves, but the recognition accuracy for flat objects decreases
Solution Approach 1:
The patent applies asymmetric warping transformation tailored to the specific geometry of floor objects and camera position. The warping parameters are calculated based on the object's detected position, distance, and the camera's viewing angle, creating a customized correction that compensates for the low-angle perspective distortion specific to floor-level capture
Solution Approach 2:
The patent introduces geometric transformation (warping) as an additional processing dimension to correct the distorted perspective. By applying 2D spatial transformation to the object region, the system recovers the object's true geometry from the distorted low-angle view, enabling accurate recognition of flat floor objects
3Measurement precision
If high-resolution cameras are used to improve recognition, then the image quality improves, but the computation amount and time increase
Solution Approach 1:
The patent extracts only the necessary object regions from the full image for warping and detailed processing. By detecting objects first and then applying computationally intensive warping operations only to these specific regions rather than the entire high-resolution image, the system maintains high recognition accuracy while significantly reducing computation time and resource usage
Data Source
Figure 1~2a
Figure 2b~2d
Figure 3a
AI summary
An electronic apparatus is provided. The electronic apparatus includes: a sensor; a camera; a memory; and a processor configured to be connected to the sensor, the camera, and the memory. The memory includes an artificial intelligence model trained to identify at least one object. The processor is further configured to: detect an object based on sensing data received from the sensor; based on the detected object being identified as having a height less than a predetermined threshold value, warp an object region, including the detected object, in an image acquired through the camera based on distance information of the object region; and identify the detected object by inputting the warped object region into the artificial intelligence model.