Automatic Inspection System of Industrial Cowl Cross Assembly uUing Hough Transform

This paper realized an inspection system of the cowl cross assembly which is an automotive part using single CCD camera. Edge is extracted by Sobel operator after enhanced image quality with homomorphic filter. The noise elements are removed by labeling from the obtained edge image. The accurate center point of the merged segment can be estimated using RHT (Randomized Hough Transform). Existence of defect can be decided by comparing the CAD information of the cowl cross assembly with the estimated center position. Efficiency of the proposed method was proven by accurate inspection result for various target images at the poor environment such as the real industrial field.