Contour Extraction Using Sequential Image Passes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current contour-extraction methods in digital images are inefficient on processors with small local memory, as they require random memory accesses and struggle with complex shapes, leading to slow processing and incomplete contour detection.
Innovation Solution
The method involves making deterministic and predictable passes across the image, prefetching rows or columns into local memory, and combining contour portions in a structured manner to create a full contour, ensuring efficient memory access and accurate shape detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If random memory access is used to traverse edge pixels in any direction, then contour detection can handle arbitrary shapes, but processing speed deteriorates due to memory access delays
Solution Approach 1:
The image processing is divided into multiple sequential passes (e.g., four passes in different directions), where each pass handles a specific directional traversal. This segmentation allows the algorithm to process contours in an ordered manner rather than random access, improving memory access efficiency while maintaining the ability to detect arbitrary shape contours.
Solution Approach 2:
The algorithm performs preliminary actions by pre-processing the image to identify edge pixels and organizing them in a structured manner before the actual contour tracing. This preliminary organization enables subsequent passes to efficiently access and connect edge pixels without random memory jumps, thus improving processing speed.
2Speed
If the entire image is loaded into local memory, then processing speed improves, but memory requirements increase beyond what low-power processors can provide
Solution Approach 1:
The image data is segmented into smaller chunks that can be processed sequentially in multiple passes. Instead of loading the entire image into local memory at once, only the necessary portions are loaded during each pass, reducing the memory capacity requirement while maintaining efficient processing speed through ordered access patterns.
Solution Approach 2:
The algorithm performs preliminary actions by pre-fetching and organizing image data into a structured format that can be processed in sequential passes. This preliminary organization allows the system to work with smaller data subsets in local memory while maintaining high processing efficiency.
3Speed
If deterministic passes are made across the image, then memory access efficiency improves, but the ability to detect complex contour shapes may be compromised
Solution Approach 1:
The contour detection process is segmented into multiple directional passes (e.g., horizontal, vertical, diagonal directions). Each pass contributes to detecting different portions of the contour, and by combining results from all passes, the algorithm achieves both memory access efficiency and complete contour detection for complex shapes.
Solution Approach 2:
The algorithm adds a temporal dimension by processing the image in multiple sequential passes rather than a single random traversal. This multi-pass approach allows deterministic memory access patterns while still capturing complex contour geometries by accumulating results from different directional perspectives.
Data Source
AI summary
Contours of objects appearing in a digital image are extracted in a plurality of one-directional passes across the digital image. Each pass loads rows or columns of the image into a local memory, in the order they appear in the image, and analyzes them for the presence of portions of the full contour. The portions are then combined to create the full contour.