Abstract:
This research proposed a new method to measure human body parts size using image for computing suitable bike frame size. This method consists of three steps: 1. The process of Background subtraction for human and reference object length 2. The process of classifying body parts and reference object 3. The process of finding the reference size, body parts size and frame size. In the result into 3 experiments: l.The average accuracy of body parts size into 10 parts : Inseam, Trunk, Forearm, Arm, Thigh, Shoulder, Torso, Lower leg, sternal notch and Total height is 93.56 % and the average accuracy of frame dimension calculation of all 9 points : Seat tube length, Top tube length, Seat tube angle, stem extension, Crankarm length, Saddle height, Saddle setback, Saddle- Handlebar difference and Handlebar is 95.76 % and last result average accuracy of bike frame size in 7 bands Giant, Merida, Scott, Jamis, Trek, Cannondale and Ritchey is 98.31 %. By this propose method, We succeeded to finding proper bike frame from human body parts using image and its can be improved to have more geometry on bike frame for coverage on many of the bikes.
