Straight Line Detection via Parameter Space Segmentation

The patent segments the parameter space into multiple local regions and performs separate voting and filtering operations in each region. This allows different threshold strategies to be applied locally - higher thresholds in regions with strong peaks to maintain accuracy, and lower thresholds in regions with weak peaks to improve coverage, thereby resolving the contradiction between coverage scope and detection accuracy

The patent applies different filtering criteria and threshold values to different local regions of the parameter space based on their characteristics. Regions with strong voting peaks use stricter filtering to maintain accuracy, while regions with weak peaks use more lenient filtering to improve detection coverage, thus achieving both high accuracy and broad coverage simultaneously

By dividing the parameter space into local regions and performing independent voting and filtering, the patent enables high thresholds to be applied in specific regions without globally reducing coverage. Each region can be optimized independently, maintaining high accuracy where needed while preserving coverage in other regions

The patent applies strict filtering partially - only in regions where strong voting peaks indicate high-confidence detections. In regions with weaker signals, more lenient filtering is applied to ensure coverage, thus achieving high accuracy for certain detections without sacrificing overall coverage scope