Abstract A model of a charge simulation retina was developed to distinguish different shapes and sizes. The retina consisted of sensory and work cells. Signals at work cells were computed based on images generated by different 3-D shapes and sizes, arbitrarily located in the field of view of the retina, and the stimuli applied on sensory cells. Using neural networks and based on the signals, overall classification rates of 73% for both shape and size were obtained. Object displacement affected the performance of the retina, especially for locations beyond 0.1R (R is the radius of retina's base). Hence, the retina can identify to a reasonable degree different 3-D shapes and sizes, in spite of arbitrarily locating the shapes in its field of view. However, it is necessary that the centres of area of the object and the retina base be within a 10% error to minimise identification errors.
[1]
E. Rosenblueth,et al.
Ground motion at canyons of arbitrary shape under incident sh waves
,
1979
.
[2]
C. T. Morrow,et al.
Fourier-based Separation Technique for Shape Grading of Potatoes Using Machine Vision
,
1995
.
[3]
M. J. Delwiche,et al.
Machine vision methods for defect sorting stonefruit
,
1994
.
[4]
C. T. Morrow,et al.
Grading of Mushrooms Using a Machine Vision System
,
1994
.
[5]
H. Murase,et al.
Environmental Control Strategies Based on Plant Responses Using Intelligent Machine Vision Technique
,
1995
.