In non-destructive evaluation or medical imaging, X-ray computed tomography (CT) is a well-established visualisation technique. Detection of the contours of different contrast image regions are an essential and necessary requirement (for example the defect recognition and measurement
in non-destructive evaluation). The two-step approach, reconstructing images from projections first and then detecting the contours from the reconstructed images, is time-consuming and the post-processing of a reconstructed image is often difficult. In recent years, helical cone-beam CT has
been widely used because of its faster scanning speed, efficient utilisation of the X-ray dose and a natural manner to scan long objects. This paper presents an algorithm for reconstructing contours directly from helical cone-beam projections based on the single-slice rebinning algorithm and
wavelet analysis. The algorithm can be briefly summarised as: (1) using the single-slice rebinning algorithm to convert the helical cone-beam projection set to a stack of fan-beam sinograms; (2) using a convolution back-projection operation to obtain the wavelet coefficients, where the convolution
filter is the combination of the fan-beam projection of a 2D wavelet with ramp filter; and (3) detecting zero crossings of the wavelet coefficients to obtain the contours. Simulation results show the usefulness and high efficiency of the presented algorithm. For some cases of noisy projection
data, the contour detection outcome by our algorithm is obviously better than that of the two-step approach. Moreover, the run-time is less than that of the two-step approach.