Though software packages developed for scanned film imagery have required rather little enhancement, there have been some subtle effects. The sensor models have required further parameters to fit the digital imagers’ geometries and deformation patterns. Some special formats and imagery with more than eight bits per band have had to be accommodated. And very large blocks of small- and medium-format imagery and their GPS/IMU metadata must be efficiently shepherded through photogrammetric processes such as triangulation. Owing to the motion of the aircraft and the limitations of gyro-stabilized camera mounts, especially in turbulent conditions, the ground footprints of the successive lines of the strip images of pushbroom line scanners are criss-crossed and jumbled. Thus the sensor model is a time-dependent one since each line has its own orientation and the software must perform an initial rectification from GPS/IMU values, resulting in strip images that can be viewed and matched. Triangulation then results in refined trajectories and more precise rectification prior to subsequent stages of the photogrammetric workflow. This process is very demanding and is not a straightforward extension of algorithms developed purely for frame images, so only a subset of the available packages offers it. Nevertheless, around one third of the high-performance imagers sold are pushbroom line scanners. Line-scanner imagery is being successfully used worldwide, from small projects to the massive US government projects that brought the ADS40 to prominence some years ago. The acceptance of this radically different approach, i.e. a line sensor producing long strip images processed in a software workflow significantly different to that used for frame images, and the recent appearance of more suppliers of imagers in this category are remarkable and indicates the practicality of the hardware and software, tempered perhaps with a sense of adventure on the part of the users. LIDAR Data densities from airborne LIDAR continues to rise, primarily due to increasing pulse and scan rates. New electronics will soon enable multiple laser pulses to be in the air at the same time. Accuracy improves too, due more to advances in GPS/IMU performance than LIDAR per se. Though fully automated software processing is still elusive, strides have been made in both the efficacy of existing interactive software and in semi-automated classification. As observed earlier, combined airborne systems to acquire both imagery and LIDAR are becoming the norm. Software to handle all this data, however, is not so straightforward. Photogrammetrically-derived DTMs have far fewer posts than the billions in some LIDAR projects, so photogrammetric software such as BAE Systems’ SOCET SET has had to be redesigned to cope. The result is powerful: viewing LIDAR point clouds against stereoscopic imagery is ideal for point classification and editing; and photogrammetric editing tools become available to the LIDAR community. Synergies emerge between photogrammetric and LIDAR developments in the area of automatic recognition and separation of buildings and trees in DTMs. The current challenge is automatic heighting from combined LIDAR and imagery: LIDAR heights, especially near discontinuities such as the edges of roofs, enable photogrammetric matching to make fewer gross errors but also to find and match the edges for detailed, precise, reliable DTMs.  Fig. 2 Very high resolution satellites for the NextView program:GeoEye-1 (left) and WorldView 1 (right). |