Imager Defect Correction Using Kernel Median Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods are inadequate for correcting multiple defective pixels, known as cluster defects, in solid state imager devices, which can significantly degrade image quality and reduce manufacturing yield.
Innovation Solution
A method that identifies defective pixels by comparing their signal values with those of neighboring pixels within a correction kernel and corrects them by substituting their values with those of non-defective neighbors, using a specific threshold-based approach to handle both single and cluster defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple neighbor replacement methods are used to correct defective pixels, then the correction process is easy to implement, but the method fails for cluster defects and excessive dark current pixels
Solution Approach 1:
The patent segments the pixel array into multiple correction kernels (e.g., 3x3, 5x5 grids) and processes defects within each kernel independently. This segmentation allows the algorithm to handle cluster defects by identifying and correcting multiple defective pixels within each kernel using median filtering and threshold-based replacement, while maintaining the simplicity of the overall correction process through modular kernel processing.
2Productivity
If multiple defective pixels (cluster defects) are present in an area, then the manufacturing yield decreases, but existing correction methods cannot effectively handle these cluster defects
Solution Approach 1:
The patent performs preliminary actions by first identifying all defective pixels within each correction kernel before executing the correction. It calculates the median value of non-defective pixels in advance and uses this median as the replacement value for all defective pixels in the kernel. This preliminary preparation enables effective correction of cluster defects while maintaining image quality and increasing manufacturing yield.
3Reliability
If stricter manufacturing tolerances are used to reduce defects, then image quality improves, but manufacturing cost increases
Solution Approach 1:
The patent implements self-service correction where the imaging device automatically identifies and corrects its own defects using the correction kernel algorithm. The system uses the median value calculated from non-defective pixels within each kernel to replace defective pixels, enabling the device to self-diagnose and self-correct without requiring stricter manufacturing tolerances or additional manual intervention, thus maintaining image quality while reducing manufacturing costs.
Data Source
AI summary
A method and apparatus that allows for the identification and correction of defective pixels and/or pixel clusters in an imaging device. The method, and implementing apparatus determines that a pixel is defective based upon a comparison of its pixel signal value with the value of neighboring pixels. In one exemplary embodiment, a pixel is defective if it is beyond a pre-determined threshold of either a high or low value from its neighboring, corrected pixels. Pixels identified as defective can be corrected using exemplary methods of the invention such as substituting a value of the defective pixel with a value of one of its non-defective neighbors.