Estimation of geometric features for three-dimensional objects using series of binary silhouettes in orthographic views

This paper deals with the estimation of three-dimension object structure and hence its geometric features such as volume, surface area, position, and orientation from its binary silhouettes using orthographic projection. Where octree and quadtree are complex and exhaustive in computer simulation, the spatial occupancy is used instead. Due to the huge data of binary image when it is saved as bit-by-bit format, binary data compression technique is used to minimize the need of large disk space. A run-length technique is used. A series of computer simulations tests the performance of this approach using a group of artificial images of different regular objects such as sphere, cylinder, and cone. Curves of percentage-accuracy against images-number are plotted to support this approach.