RCNN Object Range Analysis via Quadrant Coordinate Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis systems, such as Critical Dimension Scanning Electron Microscopes, require time-consuming pre-operations and have low error tolerance when measuring critical dimensions, especially when initial black-and-white interlaced edges differ from expected values, leading to incorrect batch measurements.
Innovation Solution
An image analysis system utilizing a region-based convolutional neural network (RCNN) model to capture and process images, calculating the center of masked objects, generating image analysis blocks from farthest coordinate points, and performing post-processing on these blocks to accurately determine object ranges, reducing calculation complexity and improving measurement accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional CDSEM methods are used with manual recipe construction, then measurement precision can be maintained, but analysis time increases significantly and operational complexity increases
Solution Approach 1:
The system performs preliminary actions by automatically detecting image characteristics and pre-processing before actual measurement. The deep learning model pre-processes images to identify patterns and features, eliminating the need for manual recipe construction and reducing pre-operation time while maintaining measurement precision.
Solution Approach 2:
The system enables self-service through automatic image analysis where the deep learning model independently performs pattern recognition, feature extraction, and measurement without requiring manual intervention. The system self-adjusts to different image types and automatically generates measurement results, reducing both time loss and operational complexity.
2Reliability
If manual recipe construction is used, then measurement reliability can be ensured, but device complexity and operational difficulty increase
Solution Approach 1:
The system replaces the mechanical manual recipe construction process with an automated deep learning-based computational system. The neural network automatically learns measurement parameters and patterns from training data, eliminating complex manual configurations while ensuring reliable measurements through consistent algorithmic processing.
Solution Approach 2:
The deep learning model acts as an intermediary between the raw image data and the measurement results. It processes images through multiple layers of computation, extracting features and generating measurements automatically, thereby reducing system complexity while maintaining reliability through standardized processing pipelines.
3Measurement precision
If entire image post-processing is performed, then measurement accuracy is maintained, but calculation time and processing complexity increase significantly
Solution Approach 1:
The system segments the image processing task by first using the deep learning model to identify and mask regions of interest, then performing post-processing only on these segmented regions rather than the entire image. This segmentation approach maintains measurement accuracy for critical dimensions while significantly reducing processing time and computational complexity.
Solution Approach 2:
The system extracts only the necessary information and regions for measurement using the deep learning model's masked output. By taking out and focusing processing only on the masked regions containing critical features, the system maintains measurement precision while eliminating unnecessary processing of the entire image, thereby improving productivity.
Data Source
AI summary
An image analysis method is provided. In the image analysis method, a to-be analyzed image is inputted into a region-based convolutional neural network (RCNN) model to obtain a masked image outputted from the RCNN. The center of a masked object in the masked image is calculated. The center is regarded as an origin of coordinate and the farthest coordinate point from the origin of coordinate in each of the four quadrants relative to the origin of coordinate are searched. The image analysis block is generated for each of the farthest coordinate points. The post-processing is performed on the image analysis blocks to obtain an object range.