Backward Projection Image Processing for Vehicle Vision Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle vision systems that use multiple cameras require substantial processing power and resources to transform camera pixel data into displayable images, leading to inefficiencies in processing capacity, RAM usage, and bus resources.
Innovation Solution
The system employs backward projection from the display pixel grid to the camera pixel grid, utilizing warping and unwarping schemes to reduce processing requirements by only handling pixels necessary for the display, thereby saving processing power and resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional forward projection methods are used to transform camera pixel data into displayable images, then complete image transformation is achieved, but processing power consumption and resource usage increase significantly
Solution Approach 1:
The patent inverts the traditional projection approach by implementing backward projection from the display pixel grid to the camera pixel grid. Instead of projecting all camera pixels forward to the display and then filtering, the system traces rays backward from each display pixel to determine which camera pixels contribute to it. This inversion fundamentally changes the processing flow to only handle necessary pixels, reducing computational load while maintaining complete image transformation quality.
Solution Approach 2:
The patent extracts only the necessary pixel data for display by using backward projection to identify and process solely those camera pixels that will actually contribute to the final display image. This extraction principle eliminates the waste of processing irrelevant pixels, directly reducing processing power consumption while ensuring that all necessary image transformation information is preserved.
2Loss of information
If all camera pixel data is processed and transformed, then complete image information is preserved, but RAM usage and bus resource consumption increase
Solution Approach 1:
The patent applies the extraction principle by using backward projection to identify and extract only those camera pixels that will be used in the final display image. By tracing rays from display pixels back to camera pixels, the system determines the minimal necessary pixel set, extracting only relevant data for processing and storage, thereby reducing RAM usage while preserving complete necessary image information.
Solution Approach 2:
The patent implements partial action by processing only the subset of camera pixels that are necessary for the display output, rather than processing all camera pixel data. This partial processing approach, enabled by backward projection, ensures that no necessary image information is lost while significantly reducing the quantity of data that needs to be stored in RAM and transmitted over bus resources.
3Manufacturing precision
If traditional image processing methods are used, then all pixel data is handled, but processing capacity and bus resources are overwhelmed
Solution Approach 1:
The patent inverts the processing direction from forward projection to backward projection, tracing rays from display pixels back to camera pixels. This inversion enables the system to identify and process only the necessary pixels for accurate image transformation, maintaining manufacturing precision while dramatically improving processing efficiency by avoiding unnecessary computations.
Solution Approach 2:
The patent extracts only the essential pixel data needed for accurate image processing by using backward projection to determine the minimal pixel set. This extraction ensures that processing accuracy is maintained for all necessary pixels while eliminating waste of processing capacity on irrelevant data, thereby improving overall productivity.
Data Source
AI summary
A vehicular vision system includes a first camera and a second camera. The system includes an electronic control unit (ECU) and a video display operable to display video images derived from frames of first image data captured by the first camera and frames of second image data captured by the second camera. Responsive to a view selection input selecting a view for the video display, the system (i) determines a first subset of pixels based on tracing rays from pixels of the video display the pixels of the first camera, and (ii) determines a second subset of pixels based on tracing rays from pixels of the video display to pixels of the second camera. The system buffers the first and second subset of pixels and generates display frames of image data using the buffered pixels. The system displays video images derived from the display frames of image data.