Accurate Playground Localisation Based On Multi-Feature Extraction And Cascade Classifier In Optical Remote Sensing Images

To address the low accuracy problem of playground detection under complex background, the accurate playground localization based on multi-feature extraction and cascade classifier is proposed in this paper. It is difficult to utilize this information to separate objects from the complex background. Therefore, we adopt multi-feature extraction method to make the playgrounds more easily to be detected. The proposed localization method is partitioned into two modules: feature extraction and classification. First, multi feature extraction method combining histogram of oriented gradients (HOG) and Haar is utilized to extract features from raw images. HOG can authentically capture the shape information, which is extracted to characterize the local region. Haar can improve the image eigenvalue calculation effectively. Afterwards, cascade classifier based on AdaBoost algorithm is adopted to classify the extracted features. Finally we conduct the experiments with our proposed methodology on a publicly accessible remote sensing images from Google Earth. The results demonstrate that the proposed framework has a better effect with achieving high levels of recall, precision and F-score compared to the state-of-the-art alternatives, without sacrificing computational soundness. What is more, the results indicate that the proposed playground 1ocalisation method has strong robustness under different complex backgrounds with high detection rate.