With the development of information networking, the application of quick response (QR) codes is more and more diverse. The recognition and application of traditional single QR code cannot meet the current demands. Therefore, a correction algorithm for multi-target QR codes is proposed. The image is binarized and the code symbolic characteristics are used to locate the QR codes. The QR codes are segmented based on those characteristics and the connected components labeling algorithm. Then, contour tracing is used to obtain the boundaries of the QR codes detection pattern. Three vertices on the detection pattern are obtained by using the geometric relationship. The fourth vertex is obtained by the relationship between the probe point and the other vertices. Finally, the inverse perspective transformation is used to correct the code shape. The proposed algorithm is implemented by C++ programming language, and the corrected QR codes are tested by Zbar. Experimental results show that the proposed algorithm can carry out the correction of multi-target QR codes quickly with high recognition rate, and can overcome the influence of different illumination and background interference, showing good robustness.