The work aims to study the detection method of crosshatch-angle to detect the internal crosshatch-angle efficiently and accurately and discard defective products, as the internal crosshatch-angle of engine cylinder bore directly affects the performance of the engine. Neighborhood operation and adaptive median filtering were adopted to enhance the crosshatch image, so as to achieve the dual effects of removing noise and preserving details. Then, the edge segmentation of the image was carried out to obtain the well-defined edge image and then the image was binarized to reduce Hough change time. Finally, a line extraction algorithm based on Hough transform and zeroing method was proposed, and the crosshatch-angle was calculated according to the extracted line features. The experimental results showed that, based on the angle detected by the proposed algorithm, the average detection error was reduced to 0.43%, compared with the traditional method and overseas universal instruments. Featured by small angle error, high accuracy and less time consumption, the proposed algorithm can replace expensive imported instruments, and it has better application value in the field of line detection and packaging barcode.