The introduction of synchrotron beamlines for high-energy X-ray diffraction raises new possibilities for texture determination of polycrystalline materials. The local texture can be mapped out in three dimensions and texture developments can be studied in situ in complicated environments. However, it is found that a full alignment of the two-dimensional detector used in many cases is impractical and that data-sets are often partially subject to geometric restrictions. Estimating the parameters of the traces of the Debye-Scherrer cones on the detector therefore becomes a concern. Moreover, the background may vary substantially on a local scale as a result of inhomogeneities in the sample environment etc. A set of image-processing tools has been employed to overcome these complications. An automatic procedure for estimating the parameters of the traces (taken as ellipses) is described, based on a combination of a circular Hough transform and nonlinear least-squares fitting. Using the estimated ellipses the background is subtracted and the intensity along the Debye-Scherrer cones is integrated by a combined fit of the local diffraction pattern. The corresponding algorithms are presented together with the necessary coordinate transform for pole-figure determination. The image-processing tools may be useful for the analysis of noisy or partial powder diffraction data-sets in general, provided flat two-dimensional detectors are used.