Multiple Thresholding Methods for Extracting & Measuring Human Brain and 3D Reconstruction

Sumijan, Sumijan and Ayu Widya Purnama, Pradani and Arlis, Syafri (2019) Multiple Thresholding Methods for Extracting & Measuring Human Brain and 3D Reconstruction. Journal of Physics: Conference Series, 1339. 012027. ISSN 1742-6588

Preview

Text Dr Sumijan (Multiple Thresholding Methods for Extracting & Measuring Human Brain and 3D Reconstruction).pdf Download (1MB) | Preview

Abstract

Thresholding changes the gray image to binary imagery and improves image quality. This study applies the multiple thresholding method to extract and calculate the area of bleeding in the human brain. 10 samples of human brain image using multiple threshold method (Otsu and hybrid thresholding). The results of the Otsu method that still have noise can be overcome by the hybrid thresholding method. The results of the original image research with images that have been processed using the multiple thresholding method yield the optimum threshold value, the method applied produces excellent image quality. Results of calculation of MSE, RMSE and PSNR. the average PSNR Otsu thresholding: an average of 60.93 dB, the average PSNR Hybrid thresholding: 59.06 dB, and the average PSNR Multiple thresholding: 56.96 dB, the average MSE Otsu thresholding: 38.27, the average MSE Hybrid thresholding 36.15 , mean MSE multiple thresholding: 34.30, mean RMSE Otsu: 9.49, mean RMSE Hybrid thresholding: 8.88, average RMSE Multiple thresholding: 7.58. The results of the calculation of the area of cerebral haemorrhage and the level of accuracy indicate a better multiple thresholding method. The results of the calculation of the bleeding area were carried out by 3D reconstruction using linear interpolation method.

Actions (login required)

View Item