Tilt Correction Toward Building Detection of Remote Sensing Images

Building detection is a crucial task in the field of remote sensing, which can facilitate urban construction planning, disaster survey, and emergency landing. However, for large-size sensing images, great majority existing works have ignored image tilt problem. This problem result partitioning buildings into separately oblique parts when images are partitioned. not beneficial to preserve semantic completeness building objects. Motivated by above fact, we first propose framework detecting objects image, particularly detection. The mainly consists two phases. In phase, correction (TC) algorithm, contains three steps: texture mapping, angle assessment, rotation. second performed with object detectors, especially deep-neural-network-based methods. Last but least, results will be inversely mapped original image. Furthermore, challenging dataset named Aerial Image Detection contributed public research. To evaluate TC method, also define an evaluation metric compute cost partition. experimental demonstrate effects proposed method