Automatic Mission Planning and Image Processing System
Unmanned Aerial Vehicle (UAV) is an aircraft that can be remotely controlled or can fly autonomously based on pre-programmed flight plans or more complex dynamic automation systems. Comparing to the general air-borne or space-borne platform, however, UAV provides an innovative approach of remote sensing that is much cheaper, safer and more flexible to be deployed in a small area, ranged from a few to tens of square kilometers. As technologies advanced fast and cost decreased dramatically in the past few years, low-cost UAVs with competitive performance are now available commercially. The existing position/attitude sensors are not only too heavy but also too expensive to be loaded on a low-cost UAV. As a result, a lot of low-cost UAV platforms are limited to qualitative applications using non-georeferenced photos.
Automatic Mission Planning and Image Processing System (AMPIPS) serves as a fast and reliable approach that enables users to plan a flight mission and generate three levels of geo-referenced products using photos/data acquired from low-cost UAV.
AMPIPS enables users to specify the flight courses to be parallel with the wind direction measured in situ to avoid cross winds. The users only need to draw the polygon of study area on the base image and specifies the in situ wind direction. The AMPIPS will make a detailed flight plan with the consideration of sufficient overlapping and turning radius, as well as automatically generate the corresponded fly file that is ready for uploading to CropCam® to fly.
The payload of a low-cost UAV is usually very limited. Thanks to the rapid advance of consumer digital cameras in the past few years. There are many non-expensive yet lightweight cameras available now that can acquire more than 10 million pixels in single photo with great quality. This makes the low-cost UAV an ideal platform to provide the standard georeferenced products on a regular basis. To serve as a metric camera, however, the consumer digital camera needs to be calibrated, particularly the correction of lens distortion for each photo. AMPIPS allows the users to select the camera type and the calibration parameters of lens distortion.
GPS/INS Geo-referenced product
The MicroPilot 2028 system onboard CropCam is able to keep a detailed log at the frequency of 4Hz during the flight mission. The log includes the real-time readings of three GPS coordinates (x, y, z), three attitude (pitch, roll, yaw), three axes acceleration, three axes rotation, current speed, servo mode, as well as the planned coordinates, planned attitude and planned speed. Based on the principle of geometric projection, the projected-photo can be generated by projecting four corners of the calibrated-photo to the ground coordinate, using the corresponded GPS/INS data recorded in the log file.
RST Geo-referenced product
According to the flight plan, the base map/image can be cut into a set of reference images. Each corresponds to one photo that would be taken by UAV during the flight . Comparing each projected-photo to its corresponding reference image, the user can select three pairs of conjugate points. An iterated calculation would be conducted to get the optimized solution of the GPS/INS data that minimize the difference between the projected coordinates and the readings from the reference image. The RST transformation can be applied to geo-reference each calibrated-photo and generate the RST-photo.
Ortho Geo-referenced product
Two approaches can be employed to generate the orthorectified images. If the digital topography model (DTM) of the study area is provided, it can be cut into a set of reference DTM, each corresponds to one photo that would be taken by UAV during the flight. With this reference DTM and three pairs of conjugate points, the calibrated-photo can be orthorectified to ortho-photo. If the overlapped areas are large enough to provide a stereo pair, the tie points would be automatically selected and denoted on the two images, respectively. The users can quickly screen them and remove those mismatched points. The rest points would enable us to get the optimized solution of the GPS/INS data for both images, as well as the DTM and ortho images. Since all ortho-photos have been geo-referenced, they can be stitched to a seamless and color-balanced mosaic using the third-party photo-stitching software.
Rapid sharing through Google Earth powered by cloud service