Object Identification via Image Scaling and Parameter Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object identification methods face accuracy issues when the difference between test and training distances is significant, requiring either a large database or low accuracy simulation to improve identification rates.
Innovation Solution
Establish a training database with photographing distances and camera parameters, and adjust test images and camera parameters to match the training database, allowing for accurate object identification without a large database or simulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large database storing many training distances is established to improve identification accuracy, then identification accuracy is improved, but database establishment time increases
Solution Approach 1:
The patent applies preliminary action by pre-establishing a training database with a limited set of representative training distances and camera parameters before actual object identification tasks. This preliminary database preparation enables the system to handle various photographing distances without requiring extensive real-time database construction, thus improving identification accuracy while avoiding excessive database establishment time.
Solution Approach 2:
The patent utilizes parameter changes by adjusting camera parameters and photographing distances to match the pre-established training database. When a test image is captured, the system modifies the camera parameters and scales the image to correspond with the training conditions, enabling accurate identification without requiring the database to contain every possible distance scenario.
2Reliability
If simulation of various sizes/angle/distance of the object is performed to improve identification rate, then identification rate is improved, but data volume to be simulated increases
Solution Approach 1:
The patent applies preliminary action by pre-simulating and storing a limited set of representative object appearances at various distances and angles in the training database. This preliminary simulation covers the most common scenarios, enabling the system to handle new objects effectively without requiring exhaustive simulation of all possible variations, thus improving identification rate while controlling data volume.
Solution Approach 2:
The patent uses copying by creating scaled versions of training images to match different photographing distances. Instead of simulating entirely new data for each distance, the system copies and resizes existing training images to represent objects at various distances, significantly reducing the data volume required while maintaining identification accuracy.
3Measurement precision
If test image is adjusted to match training image size to improve identification accuracy, then identification accuracy is improved, but image processing complexity increases
Solution Approach 1:
The patent applies parameter changes by systematically adjusting the test image parameters (size, scale) to match the training database conditions. The system calculates the appropriate scaling factor based on the ratio between test and training photographing distances, then applies uniform scaling to the test image. This parameter-based approach simplifies the processing complexity compared to more complex image transformation methods.
Data Source
AI summary
An object identification method includes: establishing a training data base including a photographing distance of a training image and a training camera parameter; in photographing a target test object, obtaining a test image, a depth image, an RGB image and a test camera parameter; and based on the training database, the depth image and the test camera parameter, adjusting the RGB image wherein the adjusted RGB image having a size equivalent to the training image of the training database.