Recognition of Non Circular Iris Pattern of the Goat by Structural, Statistical and Fourier Descriptors
Parthasarathi De, Dibyendu Ghoshal · Procedia Computer Science · 2016
The present paper has described a comparative study to find the iris pattern of the goat which has nearly rectangular or square type appearance. For detecting the structural descriptors, the deviation of the iris pattern shape and size from a standard circular (annular) shaped have been thoroughly studied. Statistical feature extraction has mainly dealt with the various types of moments e.g. – mean variance skewness and kurtosis1, 2. Fourier descriptors have been extracted by 2D Fourier Transformation of the entire data set comprising patterns. It has been found that Fourier Descriptors are not directly insensitive to possible geometrical changes of the iris location like translation, rotation and scale change occurring due to eye ball movement and blinking of the eye lids. The result shows that the structural descriptors based pattern recognition rate produce a recognition rate of 97.85% with 4.5% of false acceptance rate and 2.2% false rejection rate. The images during the study were acquired from real life with 16 megapixel camera resolution.