METODE PENCARIAN FOKUS SECARA PASIF PADA MIKROSKOP DIGITAL DENGAN STUDI KASUS SAMPEL DARAH

MOCHAMAD SYAHDANI A, Ir. Balza Achmad, M.Sc.E

2014 | Skripsi | FISIKA TEKNIK

Salah satu jenis pemeriksaan kesehatan adalah analisis darah menggunakan mikroskop. Dalam kebanyakan analisis, pengaturan fokus mikroskop masih manual sehingga lama dan rentan terhadap kesalahan manusia karena kelelahan, kelalaian, dan lain-lain. Berdasarkan permasalahan tersebut, pada tugas akhir ini dibangun sebuah metode pencarian fokus untuk mikroskop digital dengan studi kasus sampel darah. Proses penelitian meliputi pengambilan citra 10 sampel darah menggunakan mikroskop digital perbesaran 40x dan 100x pencahayaan terang serta 100x pencahayaan redup menghasilkan 30 set citra. Setiap set berisi 200 citra di mana pengambilan antar citra berjarak 1 μm. Lima responden dipilih untuk pengambilan suara memilih citra tertajam (fokus) sebagai acuan pengujian. Kemudian, sebuah program dibuat berisi 6 fungsi kuantisasi yaitu Gradien Tenenbaum, Gradien Brenner, Variansi, Variansi Ternormalisasi, Korelasi Standar Deviasi, Korelasi Auto, 4 pencarian puncak yaitu pencarian Global, Mendaki, Fibonacci, Adaptif, dan 4 daerah pencacahan yaitu Keseluruhan, 1/5, 1/9, dan Pilihan. Daerah pencacahan Pilihan tidak diuji karena bersifat pilihan bebas sehingga terdapat 72 kombinasi metode pencarian fokus. 72 kombinasi diuji melakukan pencarian fokus dengan parameter galat posisi fokus (μm) dan waktu (ms). Unjuk kerja optimal didapatkan oleh kombinasi fungsi Variansi Ternormalisasi, pencarian Adaptif dan daerah pencacahan 1/9 dengan rata-rata galat posisi fokus 1 ± 2 μm dan waktu 347 ± 110 ms per proses pencarian fokus.

One of the methods of health diagnostics is blood analysis using a microscope. In most of such analysis, the focus of the microscope is tuned manually and thus requires a longer time and is susceptible to human errors such as tiredness, carelessness, etc. Based on that problem, on this research a method of passive focus finding is developed for a digital microscope with the case study of blood samples. The research procedure includes capturing image of 10 blood samples using a digital microscope with 40x and 100x magnification with bright illumination, and 100x magnification with dark illumination resulting in 30 sets of images. Each set contains 200 images where the object is moved 1 μm for each image captured. 5 respondents are asked to vote to determine the sharpest image of each set (focus image) as a reference for characterization. Then, a program is built which provides 6 quantization function: Tenenbaum Gradient, Brenner Gradient, Variance, Normalized Variance, Standard-Deviation-Based Correlation, Auto Correlation; 4 peak searches: Global, Climbing, Fibonacci, Adaptive; and 4 sampling area: Whole, 1/5, 1/9, and Free Selection. The sampling area of Free Selection is excluded from the test so that there are 72 combinations of focus finding method. Those combinations will be tested to perform focus finding with 2 test parameters, focus error (μm) and time (ms). Optimal performance is obtained by the combination of Normalized Variance, Adaptive Search, and 1/9 sampling area with average focus error 1 ± 2 μm and average search time 347 ± 110 ms for each focus finding process.

Kata Kunci : pencarian fokus, mikroskop digital, sampel darah, fungsi kuantisasi, pencarian puncak, daerah pencacahan