Video Stream Index Compression Using Split Half-Indices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In forensic watermarking, existing methods require lengthy index values to uniquely identify data items in a stream, leading to inefficient data storage and processing, especially in streaming applications where each data instance needs log2(j) bits for identification.

Innovation Solution

The method employs a Hamiltonian Path in a directed graph to split each full index into two half-indices, each used once as a preceding and once as a following index, reducing the total index bits required by half while maintaining proximity to the data instance, and uses a recursive Depth-First Search to efficiently compute these half-index values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If full index values are used to identify each data item, then identification precision is maintained, but storage space and transmission bandwidth are wasted

Engineering Contradiction:
Improveidentification precisionVSAvoidstorage space
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides each full index value into two half-index values, where each half-index contains only part of the identification information. The first half-index is placed before the data item and the second half-index is placed after the data item, together they uniquely identify the data item without requiring the complete index value at any single location, thus reducing storage space while maintaining identification precision.

Inventive Principle:
Principle #1Segmentation

2Quantity of substance

If index values are shortened to save storage space, then storage efficiency improves, but identification reliability deteriorates

Engineering Contradiction:
Improvestorage efficiencyVSAvoididentification reliability
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent combines two half-index values (first half-index and second half-index) to form the complete identification information for a data item. Each half-index alone is insufficient for identification, but when combined, they provide reliable and unique identification, thus maintaining identification reliability while using fewer bits at each location.

Inventive Principle:
Principle #5Merging (Combining)

3Ease of operation

If conventional indexing methods are used, then implementation simplicity is maintained, but resource utilization efficiency deteriorates

Engineering Contradiction:
Improveimplementation simplicityVSAvoidresource utilization efficiency
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The patent segments the index structure into two parts distributed around the data item, which reduces the number of bits that need to be stored and transmitted at any single location. This segmentation improves resource utilization efficiency by reducing storage requirements and bandwidth consumption, while the overall indexing mechanism remains conceptually simple and easy to implement.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3443674B1Compressing indices in a video stream
Publication Date: 2020.06.17 CISCO TECHNOLOGY INC
  • EP3443674B1 patent drawingFigure 1~2
  • EP3443674B1 patent drawingFigure 3
  • EP3443674B1 patent drawingFigure 4

AI summary

In one embodiment a system, apparatus, and method for optimizing index value lengths when indexing data items in an array of data items is described, the method including producing, at a first processor, an ordered series of index values, sending the ordered series of index values to an indexing processor, receiving, at the indexing processor, a data object including the array of data items, associating, at the indexing processor, a first part of one of the index values with a first one data item of the array of data items, associating, at the indexing processor, a second part of the one of the index values with a next one data item of the array of data items, repeating the steps of associating a first part of one of the index values and associating a second part of the one of the index values until all of the data items in the array of data items are indexed.