An improved Hough Transform and its realization on FPGA

Hough Transform is a powerful and robust global image processing tool for line detection in picture. However, the traditional Hough Transform is complex because of the calculation of trigonometric functions and multiplication. To overcome this disadvantage, a novel algorithm using simple shift and addition will be discussed in this thesis to replace the complex trigonometric functions and multiplication. Meanwhile, a deep pipelined and parallel realization of this new algorithm on FPGA is also designed. In addition, a new threshold method based on the peak extraction is proposed to eliminate the fake lines produced by the traditional Hough Transform. Experiment results are also presented to prove the efficiency of the proposed method.

[1]  Manuel Hernandez Calviño,et al.  FPGA Implementation of the Generalized Hough Transform , 2009, 2009 International Conference on Reconfigurable Computing and FPGAs.

[2]  I. Martinez,et al.  VHDL described finger tracking system for real-time human-machine interaction , 2008, 2008 International Conference on Signals and Electronic Systems.

[3]  C. Hollitt.  Reduction of computational complexity of Hough transforms using a convolution approach , 2009, 2009 24th International Conference Image and Vision Computing New Zealand.

[4]  Jae Wook Jeon,et al.  A real-time finite line detection system based on FPGA , 2008, 2008 6th IEEE International Conference on Industrial Informatics.

[5]  Thambipillai Srikanthan,et al.  Parallelizing the Hough Transform Computation , 2008, IEEE Signal Processing Letters.