Identification of Beef Cattle by Analyzing Images of their Muzzle Patterns Lifted on Paper

摘要

Muzzle patterns of cattle are uneven features of their skin surface. They are different from each other as like fingerprints of human. For this reason they can be used to identify cattle. In a registration of beef cattle for breeding and marketing in Japan, the muzzle pattern is lifted on a piece of white paper in black ink within an age of 4 months and in an advanced registration at over 14- month old. However it takes a lot of time and labor to get the muzzle patterns and needs a skill by human eyes to distinguish them. In this study, an image analysis of the lifted muzzle patterns was examined to identify beef cattle automatically. Analysis proceeded from 170 head of purebred and crossbred beef cattle based on 8 purebreds, 43 in growing and 127 in finishing and breeding. After restraining of the cattle with a rope in a headgage, the muzzle pattern was lifted on the paper three times. The lifted muzzle patterns were transformed into the images with a digital scanner. From the 170 databases, 43 were sampled randomly. Based on the minimum length between the nostril borders of the sample, a square area for an image analysis was determined. A feature of the lifted muzzle pattern may locate in articulately spreading grooves colored in white or the convexes in black surrounded by the grooves. Applying the image analysis techniques of filtering, binary transforming, and thinning to the grooves in the image area, the pixels equivalent to the joints of the grooves were extracted as a key feature. The 13 samples failed to extract the joint pixels due to their blurred images. The rest were all identified using the 3 numerical indices of the numbers of them, the accumulated graphical of them, and the accumulated number of the pixels for the thinned grooves. Among the rest, 20 animals were identified as itself.