The Study of the Object Depth Estimation Based on Point Pair Features on Images

Using dual cameras to grab the object's image and then estimating the distance of this object is a useful technique. However, the feature of the object is the key role, while identifying the conjugate pair pixels of this object between the left and right images. In this study, we projected laser light on the object to create obvious features, and then apply the oFAST algorithm to extract feature points. Furthermore, ORB algorithm was applied to identify the conjugate pair pixels from all feature points on the left and right images, respectively. Finally, an epipolar plane was applied to estimate the distance between the laser point and camera lens. According to the experimental results, the error of height of the object is 2.5%. Consequently, the advance study to extract more valuable feature pixels or conjugate pairs pixel is an interesting topic.