This study presents an algorithm for image object recognition based on multi-view information fusion to deal with issues like misidentification and missed detection, due to factors such as occlusion. This algorithm proposes the EYOLO model to effectively detect the target objects in images. Then, the ORB and FLANN algorithms are employed to rapidly extract and match feature points. Meanwhile, the Random Sample Consensus algorithm is used for the fine matching of feature point pairs to obtain the optimal coordinate transformation matrix, thus achieving multi-object association. Multi-view fusion is applied to the association results to adjust the confidence values of the objects based on comprehensive association outcomes. Finally, a voting-based decision mechanism is adopted to determine the final category of the objects. The experimental results on the self-constructed dataset indicate that the proposed method improves the average precision from 81.9% (achieved by the original YOLO model) to 88.1%. On this basis, the object recognition accuracy can be improved to 98.1% through multi-view object association and fusion, which verifies the effectiveness of the proposed method.