Media Feature Comparison Using Segment Similarity Matrices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing media feature comparison methods are inefficient and inaccurate, leading to high computational and storage resource consumption.
Innovation Solution
A method and device for comparing media features by obtaining media feature sequences, determining unit similarities using cosine or Hamming distances, and constructing a similarity matrix to assess the similarity between media objects, with the option to identify matching segments through straight lines in the matrix.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing media feature comparison methods are used, then media comparison can be performed, but the accuracy and efficiency are poor, causing huge consumption of computing resources and storage resources
Solution Approach 1:
The patent segments media feature sequences into multiple feature units and compares them in a hierarchical manner. First, feature units are extracted from media sequences, then similarity is calculated between corresponding feature units, and finally overall media similarity is determined based on feature unit similarities. This segmentation approach improves comparison accuracy while reducing computational resource consumption by breaking down complex media comparison into manageable feature unit comparisons.
2Productivity
If existing media feature comparison methods are used, then media comparison can be performed, but the efficiency is poor, causing huge consumption of storage resources
Solution Approach 1:
The patent extracts essential feature units from media sequences, storing only these extracted features rather than complete media data. The similarity comparison is performed based on these extracted feature units, which significantly reduces storage resource requirements while maintaining comparison efficiency. The system stores feature unit representations (such as audio fingerprints or visual features) instead of entire media files, enabling efficient storage and rapid comparison.
Data Source
AI summary
The disclosure is related to a method and device for comparing media features, the method comprising: obtaining first media feature sequences of a first media object and second media feature sequences of a second media object, the first media feature sequence comprises a plurality of first media feature units arranged in sequence, and the second media feature sequence comprises a plurality of second media feature units arranged in sequence; determining unit similarities between the first media feature units and the second media feature units; determining a similarity matrix between the first media feature sequences and the second media feature sequences according to the unit similarities; determining a similarity of the first media object and the second media object according to the similarity matrix.