Industrial Defect Detection Image Scaling Method

The patent segments the image processing task into multiple stages: initial low-resolution filtering to eliminate obvious defective products, followed by detailed inspection only of products that pass the first stage. This multi-stage segmentation approach maintains detection accuracy for critical defects while significantly improving overall processing speed by reducing the number of high-resolution images that require intensive computation.

The patent applies different processing qualities to different parts of the workflow: low-resolution processing for initial screening and high-resolution processing only where needed for final verification. This local quality differentiation ensures that computational resources are concentrated on images that require detailed analysis, thereby balancing accuracy requirements with processing speed constraints.

The patent applies partial action by performing comprehensive high-resolution processing only on a subset of images that require detailed inspection, while applying simpler low-resolution processing to the majority of images. This approach achieves sufficient processing efficiency for critical cases without the excessive complexity and cost of applying high computational power to all images uniformly.

The patent dynamically changes processing parameters (resolution levels, computational intensity) based on the specific requirements of each inspection stage and individual product characteristics. By adjusting parameters rather than maintaining fixed high computational settings, the system achieves improved processing efficiency where needed while avoiding the sustained complexity and cost associated with always-maximum computational power deployment.