This paper presents a point pattern matching algorithm based on local relative shape context and Q-spectra of line graph.A line graph is constructed for each point

and the spectrum decomposition is performed on the signless Laplacian matrix of line graph.The eigenvalues(Q-spectra) obtained from the spectrum decomposition are used to represent the point's feature

and the initial matching probability is calculated.Local relative shape context is defined to compute the similarity distance between any two points.Q-spectra method is combined with local relative shape context via a probabilistic relaxation approach to get the matching result.Experimental results indicate the effectiveness and feasibility of the proposed algorithm.