Sains Malaysiana 49(5)(2020): 1137-1144

http://dx.doi.org/10.17576/jsm-2020-4905-19

Circular Track Finding using Hough Transform with Discretized Polling

(Penemuan Jejak Bulatan menggunakan Transformasi Hough dengan Peninjauan Diskret)

KHASMIDATUL AKMA MOHAMMAD KAMAL AZMI*, WAN AHMAD TAJUDDIN WAN ABDULLAH, AU DIYA FATIHAH & MOHAMMAD SIDDIQ

National Centre for Particle Physics, University of Malaya, 50603 Kuala Lumpur, Federal Territory, Malaysia

Received: 20 September 2019/Accepted: 23 January 2020

ABSTRACT

The objective of this paper was to show the results of single-parameter track finding based on the Hough transform method with discretized hits in an arbitrarily-sized Hough space. Each arbitrarily-sized Hough space was used to identify the most optimal hits with respect to a simulation-generated track. These hits will be useful in future studies involving multiple tracks as an identification of single-track potential size which will decrease the computational steps required to identify potential hits in multiple-track studies. These steps are well established in track finding research. However, the discretized method has not been applied fully because of uncertainty in identifying true hits in the Hough space. We have observed that by selecting the optimal discretized size, we can significantly improve the identification of true hits as it reduces the number of unrelated hits. We show that these steps are a more insightful technique compared to the traditional clustering technique by comparing our results to the K-Mean nearest neighbour method.

Keywords: Accumulator; Hough transform; K-Mean; optimal peak; track finding

ABSTRAK

Objektif kajian ini adalah untuk memperlihatkan hasil penemuan jejak satu-parameter berdasarkan kaedah transformasi Hough dengan hit yang diskret dalam ruang Hough dalam beberapa saiz tertentu. Setiap ruang Hough dengan saiz berbeza digunakan untuk mengenal pasti hits paling optimum berkenaan dengan jejak yang dihasilkan simulasi. Kajian hit ini akan berguna dalam kajian masa depan yang melibatkan pelbagai jejak sebagai pengenalan ukuran potensi satu hit yang akan mengurangkan langkah pengiraan yang diperlukan untuk mengenal pasti hits yang berpotensi dalam jejak berganda. Langkah ini dihasilkan dalam penyelidikan mencari penyelesaian. Walau bagaimanapun, kaedah diskret belum digunakan sepenuhnya kerana ketidakpastian dalam mengenal pasti hits sebenar di dalam ruang Hough. Kami telah melihat bahawa dengan memilih saiz diskret yang optimum, kami dapat meningkatkan pengenalan hits sebenar dengan ketara berbanding jumlah hit yang tidak berkaitan. Kami menunjukkan bahawa langkah ini adalah teknik yang lebih mendalam berbanding teknik berkelompok tradisi dengan membandingkan hasil kami dengan kaedah jiran terdekat Min K.

Kata kunci: Carian jejak; Min K; pengumpulan; puncak optimal; transformasi Hough

REFERENCES

Ahmed, S.R.A., Al Barazanchi, I., Jaaz, Z.A. & Abdulshaheed, H.R. 2019. Clustering algorithms subjected to K-mean and gaussian mixture model on multidimensional data set. Periodicals of Engineering and Natural Sciences 7(2): 448-457.

Alexiev, K. 2000. Implementation of hough transform as track detector. In Proceedings of the Third International Conference on Information Fusion. Volume 2. pp. THC4-11.

Bhattacharya, P., Liu, H., Rosenfeld, A. & Thompson, S. 2000. Hough-transform detection of lines in 3-D space. Pattern Recognition Letters 21(9): 843-849.

Imad Dbbura. 2018. K-means clustering-algorithm, applications, evaluation methods, and drawbacks. https://towardsdatascience.com/k-means-clustering-algorithm-applications-evaluation-methods-and-drawbacks-aa03e644b48a.

Krahe, J. & Pousset, P. 1988. The detection of parallel straight lines with the application of the Hough transform. 9th International Conference on Pattern Recognition, Rome, Italy.

Hough, P.V.C. 1962. A method and means for recognizing complex patterns. U.S. Patent: 3,069,654.

Priyanka Mukhopadhyay & Bidyut Baran Chaudhuri 2015. A survey of Hough Transform. Pattern Recognition 48(3): 993-1010.

*Corresponding author; email: khasmidatul@siswa.um.edu.my