<FONT face=Verdana>In this paper

we propose a novel rapid detector which is used to denoise and recognize edges for 2D scan line in range image. Based on silhouette analysis in a scan line

a few coefficients are defined to characterize large noise and edge points

and several determinative criteria are stipulated for rapid denoising and edge etection. Because all the coefficients used are distances or their ratio

the proposed method is inherently invariant to translation and rotation. The threshold set is trained by selecting some feature point candidates. By merging parameter attributes calculation

redundant computation is decreased to the lowest in an integral mode algorithm. Based on real data experiments

computational complexity

time expenditure

accuracy and stability of the method are evaluated in detail. The proposed detector under integral mode can process a scan line within 0.15 ms on an ordinary PC. The results indicate that the proposed method has excellent performance on efficiency

accuracy

robustness and scalability

and is qualified for real-time processing in stringent industrial occasions.