Mobile Image Processing Normal Map Determination via Depth Coordinate Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods face challenges in determining accurate normal maps for video frames, especially in mobile terminal scenarios, due to the difficulty in obtaining high-quality paired normal data and the limitations of deep learning models in terms of hardware constraints and efficiency.
Innovation Solution
A method and apparatus for image processing that determine a normal map based on a mobile terminal, involving the steps of obtaining a video frame, determining a target normal map, calculating lighting intensity information based on the normal map and light source attributes, and updating display information to create a target video frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning model-based reasoning is used to determine normal maps, then the accuracy of normal map determination is improved, but the time consumption increases and hardware constraints limit deployment
Solution Approach 1:
The patent creates a mapping relationship between depth information and normal map information through coordinate transformation. Instead of using complex deep learning models, it copies and transforms depth data into normal map data through mathematical relationships, achieving accurate normal map determination without heavy model computation.
Solution Approach 2:
The patent replaces the mechanical deep learning model system with a mathematical transformation system. By using coordinate transformation equations to convert depth information into normal map information, it eliminates the need for deploying complex neural networks on mobile hardware, significantly reducing computation time and resource requirements.
2Measurement precision
If deep learning model-based reasoning is used to determine normal maps, then the accuracy of normal map determination is improved, but the device complexity and hardware requirements increase
Solution Approach 1:
The patent creates a mapping relationship between depth information and normal map information through coordinate transformation. Instead of using complex deep learning models, it copies and transforms depth data into normal map data through mathematical relationships, achieving accurate normal map determination without heavy model computation.
Solution Approach 2:
The patent replaces the mechanical deep learning model system with a mathematical transformation system. By using coordinate transformation equations to convert depth information into normal map information, it eliminates the need for deploying complex neural networks on mobile hardware, significantly reducing computation time and resource requirements.
3Productivity
If low input resolution and small model size are set to accelerate reasoning, then the reasoning speed is improved, but the quality of output results deteriorates
Solution Approach 1:
The patent creates a mapping relationship between depth information and normal map information through coordinate transformation. Instead of using complex deep learning models, it copies and transforms depth data into normal map data through mathematical relationships, achieving accurate normal map determination without heavy model computation.
4Measurement precision
If paired normal data is collected using a camera with depth information, then the training data quality is improved, but the difficulty of data collection increases
Solution Approach 1:
The patent creates a mapping relationship between depth information and normal map information through coordinate transformation. Instead of using complex deep learning models, it copies and transforms depth data into normal map data through mathematical relationships, achieving accurate normal map determination without heavy model computation.
Data Source
AI summary
The disclosure provides a method and apparatus of image processing, an electronic device, and a storage medium. The method of image processing includes: obtaining a video frame to be processed and determining a target normal map of the video frame to be processed; determining, based on the target normal map and preset attribute information of a light source, target lighting intensity information of at least one pixel of the video frame to be processed; and determining display information of the at least one pixel based on the target lighting intensity information of the at least one pixel, and determining, based on the display information, a target video frame corresponding to the video frame to be processed.