Adaptive Modified PCA for Face Recognition

This paper presents a novel hybrid method for extracting license plates and recognizing characters from low-quality videos using morphological operations and Adaboost algorithm. First of all, the hybrid method uses the Adaboost algorithm for training a detector to detect license plates. This algorithm works well to detect license plates having lower intensities but fails to detect license plates if they are skewed. Thus, we use a morphology-based scheme to detect inclined license plates. The morphology-based scheme extracts important contrast features for searching possible license plate candidates. The contrast feature is robust to lighting changes and invariant to different transformations. The hybrid method can avoid the significant growth of training samples for training the detector to detect any oriented license plates. Then, a new segmentation method is proposed for character segmentation and recognition. Even though lower-quality video frames are handed, our method still performs very well to recognize desired license plates. The proposed technique can locate and recognize multiple plates in real time even if they have different orientations or lower intensities. Experimental results show that the proposed method improves the state-of-the-art work in terms of effectiveness and robustness for license plate recognition in low resolution and low quality source.