Specimen identification technique of melanophiniscus species through digital image processing

摘要

There is a particular interest and a broad consensus in developing reliable, fast, easy to use and inexpensive specimen identification methods that reduce stress and injuries produced by traditional marking methods. Picture based identification methods use unique natural tags or spots to identify specimens. This works presents some Digital Image Processing techniques for the identification of melanophryniscus (Sapito de las Sierras). For this purpose the spots of the ventral region are used. To accomplish the goal of comparing two given samples, it is necessary to define a region of interest, project it to a normalized view, segment the spots, calculate the area of each spot and the division by regions for the generation of the characteristic vector and finally calculate the Euclidean distance between the characteristic vectors of the samples. Positive results were obtained in the analysis on pictures taken at field trips. At last, some improvements can be done in order to compare the new sample only with the samples that there is possibility of being of the same specimen and not with all the samples of the Database.