This software implements the motion pathway of the 3D FORMOTION model.
How do spatially disjoint and ambiguous local motion signals in multiple directions generate coherent and unambiguous representations of object motion? Various motion percepts, starting with those of Duncker and Johansson, obey a rule of vector decomposition, whereby global motion appears to be subtracted from the true motion path of localized stimulus components. Then objects and their parts are seen moving relative to a common reference frame. A neural model predicts how vector decomposition results from multiple-scale and multiple-depth interactions within and between the form and motion processing streams in V1-V2 and V1-MST, which include form grouping, form-to-motion capture, figure-ground separation, and object motion capture mechanisms. These mechanisms solve the aperture problem, group spatially disjoint moving object parts via illusory contours, and capture object motion direction signals on real and illusory contours. Inter-depth directional inhibition causes a vector decomposition whereby motion directions of a moving frame at a nearer depth suppress these directions at a farther depth and cause a peak shift in the perceived directions of object parts relative to the frame.
The model is implemented in C++. The JAMA/TNT library is used for computation of the singular value decomposition, and FFTW is used for fast convolution of non-separable kernels. The code was developed for Windows 64 bits and comes with a Microsoft Visual Studio Solution File (formotion.sln). Input to the model should be encoded in XML format. A sample input is provided (divita_120_120_7_40_2_M1_0.5_1_1__10_10.xml). A pre-compiled executable is also provided (formotion.exe). Documentation can be generated using DOxygen. The program can be run by typing "formotion.exe -p plist.txt -i inputs.txt -s" at the DOS command line.
PC with Windows 64 bits
Public domain software