The FRIT-Based Image Enhancement Algorithm for Faint Pavement Cracks

摘要

Due to various circumstances, such as non-uniform lighting, weather conditions, and different pavement materials on surfaces, the pavement cracks are hardly distinguishable by eye in a complicated background. In this paper, a novel algorithm for the enhancement of faint cracks was proposed based on the finite ridgelet transform (FRIT). Furthermore we proposed the improvements on the initial design of the FRIT in order to have better energy compaction and to reduce the "wrap-around" effect. The FRIT approach, with its "built-in" linear geometrical structure, provided a more elegant way of enhancing the linear singularities and suppressing the noise. In general, the pavement cracks show the curved feature as a whole. A smooth partition method was proposed to divide the pavement crack image into localised windows where the crack looks straight. Then, we could apply the FRIT-based enhancement method in each window. Experimental results show that the new algorithm outperforms wavelets in enhancing the low-quality images with linear singularities.