Variable Hough transform for high-resolution line detection

This paper proposes a new high-resolution line detection method called the Variable Hough Transform. The Hough Transform is a method of extracting lines from an image. A line is transformed into a peak on the parameter space. However, the performance of the Hough Transform depends on the quantization on the image and the parameter space. We designed a variable filter based on the size and distribution of the quantization error, and applied it to the parameter space to improve the ability and accuracy of the line detection. Statistical experiments also show that the filter can improve the extraction ability and accuracy.