OCR String Hash Trees for Lower-Complexity Text Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing image recognition methods require significant computational resources and efficiency due to the need for comparing and storing intermediate strings and probability values, leading to high calculation complexity.
Innovation Solution
A picture recognition method utilizing hash values and tree structures to store and manage strings during text recognition, reducing the need for direct string comparisons by using mapping relationships between probability values, hash values, and strings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If intermediate strings and probability values are stored using traditional comparison methods, then text recognition can be performed, but the calculation amount becomes extremely large and storage efficiency becomes low
Solution Approach 1:
The patent uses hash values as copies of the actual strings. Instead of storing and comparing complete intermediate strings, the system stores only their hash values. This copy mechanism allows the system to maintain text recognition accuracy while dramatically reducing calculation complexity, as hash value comparison is much more efficient than string comparison.
Solution Approach 2:
The patent transforms the representation parameter of strings from the original character sequence to a hash value. This parameter change converts the complex operation of string comparison into a simpler hash value comparison, directly addressing the contradiction between maintaining recognition reliability and reducing calculation complexity.
2Reliability
If all intermediate strings are stored and compared to ensure accurate recognition, then recognition accuracy is maintained, but the storage efficiency becomes low
Solution Approach 1:
The patent replaces the storage of complete string data with storage of hash value copies. This approach maintains the ability to identify and track intermediate strings throughout the recognition process while occupying minimal storage space, thus improving storage efficiency without sacrificing recognition accuracy.
Solution Approach 2:
The patent extracts only the essential identifying feature of each intermediate string (its hash value) and stores that instead of the entire string. This extraction mechanism allows the system to maintain tracking capability for accuracy while removing redundant data, thereby improving storage efficiency.
Data Source
AI summary
The present disclosure discloses a picture recognition method performed by a computing device. The method includes: obtaining a recognized current string and a hash value of the current string during text recognition for an acquired image through an optical character recognition model, and storing the current string and the hash value of the current string into a first preset tree structure and a second preset tree structure; predicting a new probability value of the current string at a next moment, to obtain an extended string set; obtaining N strings with the highest probability value according to the hash value stored in the second preset tree structure and retaining the N strings; and using the N strings as the current strings, repeating the foregoing steps until recognition of all acquired images is completed, and obtaining a string with the highest probability value as a final recognition result.