Pengembangan Proses Analisis Citra Podoskop Semiotomatis Berbasis Python untuk Ekstraksi Area Kontak Plantar dan Arch Index

Evelyn Alicia Silaban, Prof. Dr.Eng. Ir. Herianto, S.T., M.Eng., IPU., ASEAN Eng.

2026 | Skripsi | TEKNIK INDUSTRI

Podoskop konvensional dapat menampilkan area kontak telapak kaki secara langsung, tetapi hasil pemeriksaannya umumnya masih dinilai secara visual sehingga belum menghasilkan data numerik yang objektif dan terdokumentasi. Sementara itu, sistem pengukuran tekanan plantar membutuhkan perangkat khusus dengan biaya dan aksesibilitas yang relatif tinggi. Penelitian bertujuan mengembangkan proses analisis citra podoskop semiotomatis berbasis Python untuk menghasilkan parameter citra telapak kaki secara numerik dan visual. Penelitian juga mengevaluasi kesesuaian hasil pengukuran area kontak plantar dengan metode manual menggunakan Figma dan ImageJ.

Penelitian menggunakan kerangka design thinking yang mencakup identifikasi kebutuhan melalui kajian literatur dan wawancara ahli, penentuan keluaran, pengembangan prototipe, evaluasi, serta penerapan proses. Tahapan analisis citra meliputi inventarisasi data, prapemrosesan citra, kalibrasi skala menggunakan grid 1 × 1 cm, penentuan region of interest, segmentasi awal, pengendalian kualitas interaktif, dan ekstraksi parameter. Proses diterapkan pada 150 citra podoskop dari lima mahasiswi Fakultas Teknik Universitas Gadjah Mada dalam kondisi awal serta penggunaan tas ransel, tas selempang, dan tas jinjing dengan tingkat beban 5%, 10%, dan 15?ri berat badan.

Hasil penelitian menunjukkan bahwa proses yang dikembangkan dapat menghasilkan area kontak plantar total, area kaki kanan dan kiri, area regional forefoot, midfoot, dan rearfoot, serta arch index. Keluaran visual meliputi citra segmentasi, binary mask, pembagian area regional, dan heatmap intensitas optikal relatif. Perbandingan lima citra kondisi awal menunjukkan bahwa seluruh pengukuran area kaki kiri dan kanan memiliki galat persentase di bawah 10%. Galat persentase rerata masing-masing sebesar 1,7?n 2,3%, dengan galat maksimum sebesar 4,4%. Hasil tersebut menunjukkan bahwa proses analisis citra yang dikembangkan mampu menghasilkan keluaran yang terukur dan terdokumentasi, tetapi tetap memerlukan pengendalian kualitas manual dan belum ditujukan untuk diagnosis klinis.

A conventional podoscope can directly display the plantar contact area; however, its examination results are generally assessed visually and therefore do not provide objective and documented numerical data. Meanwhile, plantar pressure measurement systems require specialized equipment with relatively high costs and limited accessibility. This study aimed to develop a semi-automatic Python-based podoscope image analysis process to generate numerical and visual foot image parameters. The study also evaluated the agreement of plantar contact area measurements with a manual method using Figma and ImageJ.

The study employed a design thinking framework involving needs identification through a literature review and expert interview, output determination, prototype development, evaluation, and process implementation. The image analysis stages included data inventory, image preprocessing, scale calibration using a 1 × 1 cm grid, region of interest determination, initial segmentation, interactive quality control, and parameter extraction. The process was applied to 150 podoscope images obtained from five female students at the Faculty of Engineering, Universitas Gadjah Mada. The images represented baseline conditions and the use of backpacks, sling bags, and tote bags at load levels of 5%, 10%, and 15% of body weight.

The results showed that the developed process could generate total plantar contact area, left and right plantar contact areas, regional areas of the forefoot, midfoot, and rearfoot, as well as the arch index. The visual outputs included segmented images, binary masks, regional area divisions, and relative optical intensity heatmaps. A comparison of five baseline images showed that all left and right foot area measurements had percentage errors below 10%. The mean percentage errors were 1.7% and 2.3%, respectively, with a maximum error of 4.4%. These findings indicate that the developed image analysis process can produce measurable and documented outputs. However, the process still requires manual quality control and is not intended for clinical diagnosis.

Kata Kunci : Analisis Citra, Podoskop, Area Kontak Plantar, Arch Index, Python, Heatmap Intensitas Optikal Relatif.