Object Recognition via Precomputed Dictionary Lookup Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition methods face challenges in balancing accuracy and processing time, particularly when recognizing various objects, as they require pre-configured feature amounts and specific processing areas, making it difficult to adapt to different objects without significant system resource increases.
Innovation Solution
An information processing apparatus that acquires image data, selects the most relevant object dictionary based on matching results, determines the object, and updates the dictionary for improved recognition, allowing for online learning and adaptation to recognition targets without pre-specifying processing areas or feature amounts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a vast number of feature amounts for various combinations of positions and sizes of local areas are used in the learning to recognize various objects with high accuracy, then the recognition accuracy is improved, but the processing time increases significantly
Solution Approach 1:
The patent pre-calculates and stores recognition results for multiple feature amount combinations in a lookup table before actual object recognition occurs. When an object needs to be recognized, the system simply queries the pre-computed results based on the actual feature amounts measured, avoiding the need to perform computationally intensive recognition calculations in real-time. This resolves the contradiction by trading off memory storage for processing time, achieving high accuracy without real-time computational burden.
Solution Approach 2:
The patent creates a copy of the recognition knowledge in the form of a lookup table that stores pre-computed recognition results. Instead of performing the actual complex recognition computations during object detection, the system copies the essential recognition patterns into a data structure that can be quickly queried. This allows the system to achieve high recognition accuracy by referencing pre-computed results rather than re-computing them, significantly reducing processing time while maintaining accuracy.
2Device complexity
If pre-configured feature amounts and processing areas are used for object recognition, then the system structure is simplified, but the adaptability to different objects is reduced
Solution Approach 1:
The patent creates a universal lookup table structure that can store recognition results for multiple different object types and feature amount combinations. The same basic system architecture and lookup table mechanism can be used for recognizing any object by simply populating the table with appropriate pre-computed data for the target object class. This resolves the contradiction by providing a simple, unified system structure that maintains high adaptability through the flexibility of the lookup table content rather than through complex structural changes.
Data Source
AI summary
An information processing apparatus acquires registration image data related to an object and input image data related to an object, and matches the acquired registration image data related to the object and each of a plurality of object dictionary data items stored in a storage device. Based on the matching result, the information processing apparatus selects an object dictionary data item relevant to the object related to the registration image data from the plurality of object dictionary data items, and matches the acquired input image data related to the object and the selected object dictionary data item. Based on the matching result, the information processing apparatus determines the object related to the input image data. Based on the determination result, the information processing apparatus updates the selected object dictionary data item. Thus, object recognition is easily and highly accurately performed.