Similarity Assessment of Agricultural Machinery 3-D Model Based on Distance-Area

Abstract:

Aimed at the retrieval performance deficiency of the classic shape distribution algorithm in the field of professional CAD models, and to achieve the reuse of agricultural machinery CAD models better, a 3-D CAD model retrieval algorithm based on distance and area distributions was proposed to achieve the model similarity assessment. Firstly, hundreds of points were selected randomly on the grid of the triangular mesh surface, during which we adopted the quasirandom number generator and Halton-Sequence to generate random points to ensure the uniformity of corresponding points; then corresponding distance between two random points as well as area ratios between each surface were calculated, and the frequency of specific points was computed in the distance-area planar grid, forming a distance-area distribution matrix of a model; finally, we adopted the Manhattan to make an assessment about different matrixes to achieve the comparison of different CAD models. Area distribution is one of the essential characteristics of models. The innovation of this paper is that we put this factor and area feature into consideration. And the descriptor was distance feature with a fusion of area to implement the accurate depiction. Accuracy was higher compared with that of other methods with respect to same retrieval parts. Halton-Sequence was adopted to generate uniform random numbers and the sampling points can be reduced significantly due to this. Thus, efficiency was acceptable. Above all, test results show that the algorithm is better in the field of agricultural CAD models compared with other shape distribution algorithms for the similarity assessment.