1. Color-Texture-Structure Descriptor

Scene classification plays an important role in understanding high resolution satellite (HRS) remotely sensed imagery. For remotely sensed scenes, both color information and texture information provide the discriminative ability in classification tasks. In recent years, substantial performance gains in HRS image classification have been reported in the literature. One branch of research combines multiple complementary features based on various aspects such as texture, color and structure. Two methods are commonly used to combine these features: early fusion and late fusion. In this paper, we propose combining the two methods under a tree of regions and present a new descriptor to encode color, texture and structure features using a hierarchical structure, Color Binary Partition Tree (CBPT), which we call the CTS descriptor. Specifically, we first build the hierarchical representation of HRS imagery using the CBPT. Then we quantize the texture and color features of dense regions. Next, we analyze and extract the cooccurrence patterns of regions based on the hierarchical structure. Finally, we encode local descriptors to obtain the final CTS descriptor and test its discriminative capability using object categorization and scene classification with HRS images. The proposed descriptor contains the spectral, textural and structural information of the HRS imagery and is also robust to changes in illuminant color, scale, orientation and contrast. The experimental results demonstrate that the proposed CTS descriptor achieves competitive classification results compared with state-of-the-art algorithms. Fig.1 is the flowchart of our proposed method.

Fig.1 Flowchart of high-resolutionsatellite (HRS) image classification based on the CTS descriptor

Table 1. Classification result comparison on UC Merced dataset

Fig.2 Classification result on the Scene-TZ Dataset

2 UAV Search and Rescue System (WiSAR) and UAV inspection System

Wilderness search and rescue (WiSAR) is very important for our country for itsvast territory.In general, WiSAR is the race with time, every second counts. Search and rescue operations often need a lot of manpower and resources . The traditional rescue method is inefficient and can easily miss the gold search and rescue time. In recent years, the rapid development of Unmanned Ariel Vehicles (UAV) has made it possible for rapid search and rescue. UAV equipped with image acquisition cameras and a variety of sensors, transport the obtained video to the ground station. We presented a wilderness search and rescue (WiSAR) system based on DJI M100 and a ground station to search and rescue the survivors in wild. We combined infrared and optical target detection to increase the detection speed and accuracy and used multiple sensors to make this system can autonomous avoiding obstructions and landing on mobile platform. For further increase the Average Precision of SSD, we build a field people dataset UAV-PP and use ResNet-101 as the base net. The actual flying test have been conducted in multiple situations to verify the feasibility of ourWiSAR system. OurWiSAR system laying a solid foundation for building a more intelligent search and rescue system based on UAV.

Fig.3 WiSAR UAV System

Fig.4 Human detection results of UAV images