A Line Detection Algorithm Based on Error Propagation

The line detection by traditional Hough transform has two insufficiencies, one is computationally expensive and memory consuming, the other is the insufficiency of accuracy. The scheme adopts the principle of error propagation, analyses the error of line detection by Hough transform in detail, and obtains the factors influencing the error. The error not only depends on the noise of the image, but also depends on the distance of the line to origin. A 3-3 window is used to search the best-distinguished pixels and limit the scope transform. Basing on the best-distinguished pixels and error propagation, the algorithm can improve the accuracy of the line detection by experiment. The computational efficiency of the scheme is not very well because of analyzing the error and finding the best-distinguished pixels, but for the application of high accuracy, it works well.