ROI-Aware Bit Rate Allocation Using Multi-Operator Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video processing methods fail to effectively improve the subjective quality of videos by optimizing bit rate allocation across multiple regions of interest, leading to unsatisfactory user experience.
Innovation Solution
A method for bit rate allocation that involves obtaining multiple region of interest (ROI) detection results using different detection operators, generating a target block quantization parameter offset mask map, and performing bit rate allocation based on this mask map to enhance image and video quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple ROI detection operators are used to improve detection accuracy, then the precision of region identification improves, but the computational complexity and processing time increase
Solution Approach 1:
The patent divides the image processing into multiple independent detection operators, each targeting specific types of regions of interest. Multiple detection operators work in parallel to identify different ROI types (e.g., face detection, object detection, text detection), allowing the system to achieve high detection accuracy without requiring a single complex detection system. This segmentation approach resolves the contradiction by distributing the detection task across multiple specialized operators.
Solution Approach 2:
The patent introduces a multi-dimensional detection framework where detection results from multiple operators are integrated through a unified mask map generation process. By adding the dimension of mask map integration and priority-based region merging, the system achieves high detection accuracy while managing complexity through structured multi-dimensional processing rather than a single complex operator.
2Manufacturing precision
If more bit rate is allocated to ROI regions to improve image quality, then the subjective quality of video improves, but the overall compression efficiency decreases
Solution Approach 1:
The patent applies different quality levels to different regions of the image based on their importance. ROI regions identified through multiple detection operators receive higher bit rates and better compression quality, while non-ROI regions use lower bit rates. This local quality differentiation allows the system to improve subjective video quality in important regions without proportionally increasing the overall bit rate, thus maintaining compression efficiency.
Solution Approach 2:
The patent dynamically adjusts the quantization parameter (QP) offset based on the detected ROI regions. By changing the QP parameter locally in ROI areas versus non-ROI areas, the system optimizes the balance between image quality and compression efficiency. The QP offset map generated from multiple detection results enables fine-grained parameter control that improves quality where needed while maintaining overall compression performance.
3Device complexity
If a single detection operator is used to reduce processing complexity, then the computational load decreases, but the ability to identify multiple types of regions of interest deteriorates
Solution Approach 1:
The patent merges the results from multiple detection operators into a unified mask map through an integration process. By combining detection results from different operators (e.g., face detection mask, object detection mask, text detection mask) into a single comprehensive mask map, the system achieves versatile ROI identification capability while managing processing complexity through structured merging operations rather than handling each detection result separately throughout the entire pipeline.
Data Source
AI summary
A method for bit rate allocation, an apparatus, an electronic device and a storage medium are disclosed, which relate to the field of artificial intelligence technology, such as computer vision, image processing and video encoder. The method for bit rate allocation includes: obtaining N region of interest (ROI) detection results of an image to be processed, wherein N is a positive integer greater than one, and the N ROI detection results comprise: detection results obtained by performing ROI detection on the image to be processed using N different detection operators respectively; generating a target block quantization parameter offset mask map corresponding to the image to be processed based on the N ROI detection results; performing a bit rate allocation on each image block in the image to be processed based on the target block quantization parameter offset mask map.