Range image recognition based on Bayesian decision rule

Laser radar is able to simultaneously produce intensity image and 3D range image which provide rich geometric information for object recognition. However, back scatter noise hinders its usage for most laser radars. Range gated imaging laser radar, which can restrain back scatter noise effectively, becomes an emerging research trend. Its specific imaging system makes the recognition method of range gated imagery different from that of other laser radars. In this paper, a new recognition approach is proposed. We represent range image by a series of range slices so that the shape-matching problem is transformed into similarity measure between 2D slices. The correlation coefficient is used to measure the similarity between 2D range slices. The final decision is made by Bayesian decision rule. Experiments demonstrate that the proposed approach is efficient in object recognition of range-gated imaging laser radar. It can handle partial occlusion and noise with high recognition rate as well.