Matrix Element Association via Submatrix Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining associations between matrix elements and comparison matrix elements in camera-based systems are inefficient, leading to high memory requirements, increased computing power, and potential loss of correspondence due to frequent overwrites in correspondence tables.
Innovation Solution
The method divides matrices into submatrices assigned to correspondence tables, allowing for a writing process to store and a reading process to retrieve positions, reducing memory usage and computing power by limiting searches to smaller tables, and incorporating an offset to manage table overwrites and optimize processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single correspondence table is used to store all matrix element positions, then complete correspondence information can be maintained, but memory requirements and computing power increase significantly
Solution Approach 1:
The patent divides the matrix into multiple submatrices, each assigned to a separate correspondence table. This segmentation allows the system to maintain complete correspondence information within each submatrix while significantly reducing the memory requirements of individual tables, avoiding the need for a single large table that would consume excessive memory resources.
2Reliability
If a single correspondence table is used for all matrix elements, then all correspondences can be stored, but processing time increases due to larger search spaces
Solution Approach 1:
By segmenting the matrix into submatrices and assigning each to a separate correspondence table, the patent reduces the search space for each table. This allows faster processing and retrieval of correspondence information within each submatrix, thereby reducing overall processing time compared to searching through a single large table containing all matrix elements.
3Measurement precision
If correspondence tables are frequently updated to maintain accuracy, then correspondence precision is improved, but the likelihood of losing correspondence due to overwrites increases
Solution Approach 1:
The patent divides the correspondence storage into multiple smaller tables, each handling a specific submatrix. This segmentation reduces the frequency of overwrites in each individual table compared to a single large table, as updates are distributed across multiple tables. Consequently, correspondence precision is maintained through regular updates while the reliability of retention is improved by reducing overwrite conflicts.
Solution Approach 2:
The patent establishes an offset between the writing process and the reading process, allowing the reading process to access correspondence information before it is overwritten by new data. This preliminary action ensures that valid correspondence data is retrieved before being lost, maintaining both precision and reliability.
4Quantity of substance
If the entire matrix is processed sequentially, then memory usage is reduced, but processing speed decreases due to lack of parallelism
Solution Approach 1:
The patent divides the matrix into multiple submatrices, each assigned to a separate correspondence table. This segmentation enables parallel processing of different submatrices, as each can be handled independently by its own correspondence table. This maintains low memory usage per table while significantly improving processing speed through parallel execution, resolving the contradiction between memory efficiency and processing productivity.
Data Source
Figure 1
Figure 2
AI summary
The present invention relates to a method for determining an assignment between a matrix element (120) of a matrix (100) and a comparison matrix element (220) of a comparison matrix (200) by means of several correspondence tables (300) whose table fields are at least partially described in a writing process. The procedure comprises the following steps: splitting the matrix (100) into a plurality of submatrices (101, ... 104), each submatrix (101, ... 104) being assigned a correspondence table (300, 301 ... 304); executing a write process to write the position of a matrix element (120) of a submatrix (101, 102 ... 104) into a table cell of the assigned correspondence table (301, ... 304); and executing a read process to read the position of a matrix element (120) from a table cell of a correspondence table (300, 301 ...304) and creating an assignment between a position of a comparison matrix element (120) and a matrix element. The writing process comprises the following steps: reading a feature of a matrix element (120) of a submatrix (101, ... 104) of the matrix (100), determining an access feature from the feature of the matrix element (120), determining a table field of the associated correspondence table (300, 301, ... 304) from the access feature, and writing the position of the matrix element (120) of the submatrix (101, ... 104) of the matrix (100) into the determined table field of the associated correspondence table (300, 301, ... 304). As well as splitting the comparison matrix (200) into a plurality of comparison submatrices (201, ... 204), where each comparison submatrix (201, ... 204) is assigned one of the correspondence tables (301, ... 304). The reading process comprises the steps: reading a feature of a comparison matrix element (220) of a comparison submatrix (201, ...204) of the comparison matrix (200), determining an access characteristic from the characteristic of the comparison matrix element (220), determining a table field of the associated correspondence table (301, ... 304) using the access characteristic of the comparison matrix element (220) and reading the stored position of the matrix element (120) from the determined table field of the associated correspondence table (301, ... 304) as well as forming an assignment between the position of the matrix element (120) read from the table field of the associated correspondence table (301, ... 304) and the current position of the comparison matrix element (220) of the comparison matrix (200).