Ceramic image processing using the second curvelet transform and watershed algorithm

Ceramic materials physical properties are the results from their microstructures characteristics. So, it is very important to quantify these characteristics in an accurate way. Ceramic microscopic image processing may be divided into two main procedures: image pre-processing for noise reduction and edges enhancement to clarify the image and image segmentation for locating and detecting the objects of interest. Conventional 2D wavelet transform is separable and thus can not sparsely represent non-separable structures of the image, such as directional curves. Curvelet transform is a new extension to wavelet transform in two dimensions. The directionality feature of curvelet transform makes it a good choice for representation of curves and edges in the image. The second generation curvelet transform theory makes it understood and implemented more easily. In this paper, a new enhancement function is proposed to enhance the edges by curvelet transform. The method is also compared with two usual contrast enhancement methods, histogram equalization and wavelet based contrast enhancement. It is shown that the contrast enhancement method based on curvelet transform is superior to both histogram equalization and the method based on wavelet transform, in both enhancement of fine structures and reduction of noise in the images. This enhancement method is used to ceramic images pre-processing, and then watershed algorithm is applied to the segmentation. The grain size distributions can be obtained from segmentation images. It has been proved that this method is effective for the ceramic grain image.