Transparent object detection is a difficult problem in the field of machine vision. Transparent objects can be rapidly detected by using multi-view light field information and polar space constraints. However, such algorithms often have low efficiency due to the large number of feature points and high similarity. To solve the above problems, the adaptive density clustering method is used to filter the feature points set, and the global and local feature significance and motion consistency constraints are combined to ensure the accuracy of feature matching and improve the speed of the algorithm. Furthermore, the slope difference between the horizontal and vertical dimensions of the feature information in the polar space domain is used to realize the target detection, which reduces the fitting process and realizes the rapid detection of transparent objects. The results show that compared with other similar algorithms, the algorithm proposed in this paper can improve the operation speed by more than 5 times while ensuring the detection accuracy.