NONSTANDARD METHODS OF ANALYSIS - GRAPHIC IMAGE-PROCESSING

In the study of biological rhythms several mathematical and statistica l techniques are commonly used for time series analysis (periodograms, least squares regression, spectral analysis, etc.). Nevertheless grap hic methods are still widely used, due to their ability to show the st ructure of rhythmic patterns. Perhaps the most widely used method is t he double plotted graph, which becomes necessary in certain specific c ases (in determining the period of an individual ultradian component, or of a weak component masked by external influences). In order to inc rease the quality of graphs, for visual inspection, several methods ar e proposed to reduce background noise, and emphasize rhythmic componen ts. The computer manipulation of large sets of data facilitates the ap plication of graphic image processing methods, here described. The fol lowing procedures are exposed and discussed, in the article: averaging , linear convolution transformation, median smoothing, averaging using a rotating mask, representation of coefficient of variation, adaptati ve neighbourhood contrast enhancement, contrast transformation and max ima transformation. Recommendations to take into account when applying these procedures individually or in short sequences, are included.