BAB VI: DYNAMIC FUZZY EXPERT SYSTEM MODEL
6.2 ANALISIS DYNAMIC FUZZY EXPERT SYSTEM MODEL
6.2.1 Sistem Inferensi Fuzzy 1 (Output: Velocity)
Perlu diingat bahwa suatu blob dapat terdiri dari beberapa orang dan yang dihasilkan pada sistem inferensi fuzzy yang pertama ini bukanlah kecepatan pergerakan tiap orang pada blob tersebut, melainkan kecepatan pergerakan blob.
6.2.1.1 Penentuan Himpunan Fuzzy
Sistem inferensi fuzzy 1 ini menggunakan tiga variabel input. Berikut rinciannya:
1. X movement change, adalah perubahan titik berat blob pada absis (sumbu x). Memiliki domain [-30,30] dengan lima himpunan fuzzy (dramatically-left, slightly-left, almost-no-change, slightly-right, dramatically right), dimana posisi left-right (kiri-kanan) diukur dari sudut pandang kamera tentunya.
2. Y movement change, adalah perubahan titik berat blob pada ordinat (sumbu y). Memiliki domain [-30,30] dengan lima himpunan fuzzy (dramatically-away, slightly-away, almost-no-change, slightly-to, dramatically-to). Kata „away‟ di sini berarti gerakan blob yang menjauhi kamera, sedangkan „to‟ berarti sebaliknya.
3. Distance, merupakan koordinat sumbu y pada titik berat blob. Memiliki domain [0,336] dengan tiga himpunan fuzzy (far, medium, close). Distance dapat juga dikatakan sebagai jarak blob dari kamera.
Kembali ditegaskan bahwa dua input yang pertama serupa dengan yang tertera pada paper Uri Kartoun [2]. Perbedaannya hanyalah pada domain dari variabel input itu sendiri. Untuk dapat mengatur domain nilai yang sesuai, penulis mengamati beberapa data sampel, mulai dari data orang yang diam hingga berlari. Terutama untuk orang yang berlari dengan
kecepatan tinggi, penulis harus dapat mengukur perkiraan nilai maksimum perubahan titik berat blob berdasarkan kecepatan lari manusia normal.
Variabel output yang dihasilkan adalah velocity yang merupakan kecepatan perpindahan blob. Memiliki domain [-2,2] dengan empat himpunan fuzzy (standing, walking-slow, walking-fast, running).
Fungsi keanggotaan yang digunakan untuk setiap himpunan fuzzy masih sama, yakni fungsi keanggotaan Gauss.
6.2.1.2 Penentuan Aturan Fuzzy
Untuk membentuk aturan-aturan fuzzy yang sesuai tentunya akan sangat bergantung pada beberapa data sampel yang diberikan. Data sampel tersebut diperoleh dari beberapa frame/citra yang menggambarkan keseluruhan kemungkinan pergerakan blob (baik dari segi arah maupun kecepatan). Jika pada fase sebelumnya suatu data dapat diperoleh dari satu citra, maka pada fase ini digunakan dua buah citra pada rentang lima buah citra untuk mendapatkan data berupa perubahan titik berat blob pada sumbu x dan y. Sedangkan untuk data berupa jarak (distance) blob dari kamera cukup menggunakan data titik berat blob pada citra yang terbaru.
Dari banyaknya himpunan fuzzy yang dilibatkan, akhirnya dihasilkan 30 aturan fuzzy sebagai berikut:
[R1] IF x-movement-change is ALMOST-NO-CHANGE and y-movement-change is ALMOST-NO-CHANGE and distance is NOT-FAR THEN velocity is STANDING
[R2] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is DRAMATICALLY-AWAY THEN velocity is RUNNING
[R3] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is SLIGHTLY-AWAY and distance is FAR THEN velocity is RUNNING
[R4] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is SLIGHTLY-AWAY and distance is FAR THEN velocity is RUNNING
[R5] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is DRAMATICALLY-AWAY and distance is FAR THEN velocity is RUNNING
[R6] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is SLIGHTLY-AWAY and distance is NOT-FAR THEN velocity is WALKING-SLOW
[R7] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is SLIGHTLY-AWAY and distance is NOT-FAR THEN velocity is WALKING-FAST
[R8] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is DRAMATICALLY-AWAY and distance is NOT-FAR THEN velocity is WALKING-FAST
[R9] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is DRAMATICALLY-AWAY THEN velocity is RUNNING
[R10] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is SLIGHTLY-AWAY and distance is FAR THEN velocity is RUNNING
[R11] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is SLIGHTLY-AWAY and distance is FAR THEN velocity is RUNNING
[R12] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is DRAMATICALLY-AWAY and distance is FAR THEN velocity is RUNNING
[R13] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is SLIGHTLY-AWAY and distance is NOT-FAR THEN velocity is WALKING-SLOW
[R14] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is SLIGHTLY-AWAY and distance is NOT-FAR THEN velocity is WALKING-FAST
[R15] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is DRAMATICALLY-AWAYand distance is NOT-FAR THEN velocity is WALKING-FAST
[R16] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is DRAMATICALLY-TO THEN velocity is RUNNING
[R17] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is SLIGHTLY-TO and distance is FAR THEN velocity is RUNNING
[R18] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is SLIGHTLY-TO and distance is FAR THEN velocity is RUNNING
[R19] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is DRAMATICALLY-TO and distance is FAR THEN velocity is RUNNING
[R20] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is SLIGHTLY-TO and distance is NOT-FAR THEN velocity is WALKING-SLOW
[R21] IF x-movement-change is DRAMATICALLY-LEFT and y-movement-change is SLIGHTLY-TO and distance is NOT-FAR THEN velocity is WALKING-FAST
[R22] IF x-movement-change is SLIGHTLY-LEFT and y-movement-change is DRAMATICALLY-TO and distance is NOT-FAR THEN velocity is WALKING-FAST
[R23] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is DRAMATICALLY-TO velocity is RUNNING
[R24] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is SLIGHTLY-TO and distance is FAR THEN velocity is RUNNING
[R25] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is SLIGHTLY-TO and distance is FAR THEN velocity is RUNNING
[R26] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is DRAMATICALLY-TO and distance is FAR THEN velocity is RUNNING
[R27] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is SLIGHTLY-TO and distance is NOT-FAR THEN velocity is WALKING-SLOW
[R28] IF x-movement-change is DRAMATICALLY-RIGHT and y-movement-change is SLIGHTLY-TO and distance is NOT-FAR THEN velocity is WALKING-FAST
[R29] IF x-movement-change is SLIGHTLY-RIGHT and y-movement-change is DRAMATICALLY-TO and distance is NOT-FAR THEN velocity is WALKING-FAST
[R30] IF x-movement-change is ALMOST-NO-CHANGE and y-movement-change is ALMOST-NO-CHANGE and distance is FAR THEN velocity is WALKING-SLOW
Untuk membuat suatu aturan fuzzy, penulis mencoba meneliti data sampel pada berbagai output pengamatan secara berurut. Misal, pertama-tama penulis mencoba membuat kumpulan aturan fuzzy yang sesuai untuk membentuk output walking-slow pada berbagai kondisi input. Setelah itu, dilanjutkan dengan output-output yang lainnya.