Before discussing the IBIS format let us first consider the standard VICAR image file. The primary purpose of image files is the storage of visual information. Images are two (or sometimes three) dimensional arrays of data, with no fundamental difference between the vertical and horizontal dimensions of the file, which are generally geometric. The fact that two elements (pixels) share a common coordinate value means little, since the coordinate axes are arbitrary. The data within an image file is homogeneous -- that is, all of the elements of the array are of the same type, format, and have the same units (if any). The nomenclature of images refers to the vertical coordinate as the "line", and the horizontal coordinate as the "sample", drawing from the operation of digital image sensors operating on a scanline basis, with the individual pixels being successive samplings of the sensor within a scanline.
Both tabular and graphics formats may also be thought of as two-dimension arrays of data of varying data types, but there are some dramatic differences as well.
Tabular files have a much broader scope than images and are used in tabulating statistics, indexing, geographic information systems, tiepoint generation for image mosaics, color lookup tables (VIDS), multispectral classification, and as serving as a general-purpose spreadsheeting utility. Tabular files, therefore, require a more complex structure than image files. In the IBIS system the vertical columns of data must consist of homogeneously formatted data, and the horizontal rows may contain heterogeneous data records of varying types. The elements having the same row coordinates are strongly related to one another, and are often used as a tabulated set of attributes or statistics for a class or sampling. The elements of a column are related in that they all represent the same type of measurement of an attribute, all have the same format and data type. Neither row nor column coordinates may have much to do with geometry, and sometimes the order within a row or column may be entirely arbitrary.
Graphics files are generally used to describe polygonal regions, and consist of a list of coordinate vectors of the successive vertices of the polygons. In fact, graphics may be considered in a subclass of tabular, in that it consists solely of numerical values, while tabular may also contain text. In the tabular model, then, each row is associated with a single vertex, and the row order in a column determines the sequence of plotting the vertices. The only special values in a graphics file are when the coordinates are all zero, which indicates a "pen-up" command, rather than a geometric position.