Perbandingan Akurasi Pengukuran Posisi dan Dimensi dengan Kamera Smartphone Standar dan Penambahan Modul GNSS RTK untuk Pemodelan Objek 3D

Sadewa Fattahsatriya, Dr. Ir. Catur Aries Rokhmana, S.T., M.T., IPU.

2026 | Skripsi | TEKNIK GEODESI

Pemodelan tiga dimensi (3D) menggunakan kamera smartphone dengan metode fotogrametri jarak dekat semakin berkembang seiring meningkatnya kualitas kamera smartphone. Namun, pemodelan berbasis smartphone umumnya masih memerlukan refrensi skala seperti scale bar untuk memodelkan objek dengann geometri yang baik karena kemampuan GNSS internal smartphone yang kurang akurat. Penelitian ini bertujuan untuk membandingkan akurasi model yang dihasilkan menggunakan kamera smartphone standar berbasis scale bar dengan model yang dihasilkan menggunakan kamera smartphone yang ditambahkan dengan modul Global Navigation Satellite System (GNSS) Real Time Kinematic (RTK). Fokus penelitian meliputi perbandingan ketelitian posisi model berdasarkan Independent Check Point (ICP) serta ketelitian dimensi panjangan objek hasil pemodelan.

Penelitian dilakukan menggunakan metode fotogrametri jarak dekat dengan algoritma Structure from Motion (SfM). Akuisisi data dilakukan pada objek lapangan basket Fakultas Teknik, Universitas Gadjah Mada menggunakan smartphone Redmi 13C. Dua skenario pengambilan data dilakukan, yaitu menggunakan smartphone standar dengan bantuan scale bar dan menggunakan smartphone yang terintegrasi modul GNSS RTK melalui pendekatan GPS-assisted photo. Sebanyak 282 foto digunakan dalam proses pemodelan, dilengkapi dengan 44 titik ICP yang diukur menggunakan GNSS geodetik metode RTK-NTRIP serta 24 data pengukuran dimensi panjangan menggunakan pita ukur sebagai data validasi. Pengolahan data dilakukan menggunakan perangkat lunak Agisoft Metashape, sedangkan evaluasi akurasi dilakukan melalui perhitungan Root Mean Square Error (RMSE).

Hasil pengujian ketelitian dimensi menunjukan bahwa model yang dibangun menggunakan smartphone dengan tambahan GNSS RTK menghasilkan nilai RMSE sebesar 0.028 m, sedangkan model yang dibangun menggunakan smartphone standar dengan bantuan scale bar menghasilkan nilai RMSE sebesar 0.029 m dengan demikian selisih kedua metode sebesar 0.001 m. Pada pengujian koordinat posisi menggunakan ICP, model berbasis GNSS RTK menghasilkan nilai RMSE sebesar 0.0336 m, sedangkan model berbasis smartphone standar menghasilkan nilai RMSE sebesar 33.359 m. Hasil tersebut menunjukkan bahwa model kamera smartphone standar tetap menghasilkan nilai RMSE dimensi panjangan sebesar 0.029 m, meskipun nilai RMSE koordinat posisinya sebesar 33.359 m. Adapun model dengan penambahan modul GNSS RTK menghasilkan nilai RMSE koordinat posisi sebesar 0.0336 m dan nilai RMSE dimensi panjangan sebesar 0.028 m.

Three-dimensional (3D) modeling using smartphone cameras with close-range photogrammetry methods has continued to develop along with the improvement of smartphone camera quality. However smartphone-based modelling generally still requires scale references, such as scale bars, to produce geometrically accurate models because the positioning accuracy of the smartphone’s internal GNSS remains limited. This study aims to compare the accuracy of models generated using a standard smartphone camera with scale bars and those generated using a smartphone integrated with a Global Navigation Satellite System (GNSS) Real Time Kinematic (RTK) module. The comparison focuses on positional accuracy based on Independent Check Points (ICP) and dimensional accuracy of the resulting model.

The study was conducted using a close-range photogrammetry approach based on the Structure from Motion (SfM) method. Data acquisition was carried out at the basketball court of the Faculty of Engineering, Universitas Gadjah Mada, using a Redmi 13C smartphone. Two acquisition scenarios were implemented: (1) a standar smartphone with scale bars as scale references and (2) a smartphone integrated with a GNSS RTK module utilizing a direct georeferencing approach. A total of 282 photographs were acquired for 3D reconstruction. In additon, 44 ICP were surveyed using a geodetic GNSS with the RTK-NTRIP method, and 24 dimensional measurments using a measuring tape as validation data. Data processing was performed using Agisoft Metashape, while accuracy assessment was conducted using the Root Mean Square Error (RMSE) calcultaion.

The dimensional accuracy test results show that the model generated using a smartphone with an additional GNSS RTK module produced an RMSE value of 0.028 m, while the model generated using a standard smartphone with the aid of a scale bar produced an RMSE value of 0.029 m, resulting in a difference of 0.001 m between the two methods. In the positional coordinate accuracy test using ICP, the GNSS RTK-based module produced an RMSE value of 0.0336 m, whereas the standard smartphone-based model produced an RMSE value of 33.359 m. These results show that the standard smartphone camera still produced a dimensional RMSE value of 0.029 m despite having a positional coordinate RMSE value of 33.359 m. Meanwhile, the model with the additional of a GNSS RTK module produced a positional coordinate RMSE value of 0.0336 m and a dimensional RMSE value of 0.028 m.

Kata Kunci : Fotogrametri Jarak Dekat, GNSS RTK, Pemodelan 3D, Smartphone, Structure From Motion.