基于会聚型双目立体视觉的测距模型

Ranging model based on convergent binocular stereo vision

传统的双目视觉模型基于特征对左右画面进行匹配,但当画面中出现多个高度相似且表面特征不明显的目标时,无法通过特征进行有效匹配。针对上述问题,本文采用基于特征和空间位置的方式进行目标匹配,并根据人类双眼锁定目标时,目标始终处于视野中心位置的原理,选择舵机来控制摄像头光轴会聚于目标,建立了一种基于会聚型双目立体视觉的测距模型。该模型能够通过控制舵机使摄像头自动汇聚于目标,并实现测距功能。理论推导及实验研究验证了该模型的可靠性及一定的实用性。

The traditional binocular vision model matches the left and right images based on features,but when there are multiple highly similar objects with inconspicuous surface features in the image,effective matching by features is not possible.In view of the above problems,this paper adopts the method based on features and spatial position to match the object,and according to the principle that the target is always in the center of the field of view when the human eyes lock the object,the steering gear is selected to control the optical axis of the camera to converge on the object,and a ranging model based on convergent binocular stereo vision is proposed in this paper.The model can make the camera automatically converge on the object by controlling the servo and realize the ranging function.Through theoretical derivations and experimental studies,the reliability and certain practicability of the proposed model are verified.