Abstract: For collected barcodes images, an image analysis method with the ability of locating, segmenting, correcting and recognizing the QR code is introduced. In order to meet real-time demands and obtain the optimal thresholds under various lighting conditions, we propose an improved adaptive thresholding, which is proved with high efficiency and high stability. In order to get the four corner points of QR code in the segmentation of barcode, a new algorithm is presented to obtain three points among those corner points according to their patterns features. The fourth corner point is then obtained by employing known point. Hough transformation. Experiment shows that the algorithm can get the four corner points of QR code effectively under various acquisition situations.