Unmanned aerial vehicles (UAVs) have been recently deployedin various civilian applications such as environmentalmonitoring, aerial imaging or surveillance. Small-scaleUAVs are of special interest for first responders since theycan rather easily provide bird’s eye view images of disasterareas. In this paper we present a hybrid approach to mosaickan overview image of the area of interest given a setof individual images captured by UAVs flying at low altitude.Our approach combines metadata-based and imagebasedstitching methods in order to overcome the challengesof low-altitude, small-scale UAV deployment such as nonnadirview, inaccurate sensor data, non-planar ground surfacesand limited computing and communication resources.For the generation of the overview image we preserve georeferencingas much as possible, since this is an importantrequirement for disaster management applications. Ourmosaicking method has been implemented on our UAV systemand evaluated based on a quality metric.

/pdf/incremental-mosaicking-of-images-from-autonomous-small-scale-1sfjhy8e3x.pdf

Incremental Mosaicking of Images from Autonomous, Small-Scale UAVs

An apparatus for generating an overview image of a plurality of images comprises a storage unit and an image processor. The storage unit stores a plurality of processed images of the overview image and is able to provide the overview image containing the plurality of processed images at their assigned positions for displaying. The image processor determines feature points of a new image and compares the determined feature points of the new image with feature points of a stored processed image to identify common feature points and to obtain 3-dimensional positions of the common feature points. Further, the image processor determines common feature points located within a predefined maximum distance of relevance to a reference plane based on the 3-dimensional positions of the common feature points to identify relevant common feature points. Further, the image processor processes the new image by assigning the new image to a position in the overview image based on a comparison of an image information of each relevant common feature point of the new image with an image information of each corresponding relevant common feature point of the stored processed image without considering common feature points located beyond the predefined maximum distance of relevance to the reference plane.

Apparatus and Method for Generating an Overview Image of a Plurality of Images Using a Reference Plane

An apparatus for generating an overview image of a plurality of images includes an image preprocessor which preprocesses a new image by assigning the new image to a position in the overview image based on position information contained by meta-data of the new image. A storage unit stores a plurality of images of the overview image and provides the overview image for display. Further, the image processor receives accuracy information of the position information. The image processor determines an overlap region of the preprocessed new image and a stored image within the overview image based on the assigned positions of the preprocessed new image and of the stored image. Further, a controllable processing engine processes the preprocessed new image by re-adjusting the assigned position of the preprocessed new image based on comparing features of the overlap region of the preprocessed new image and the stored image. The controllable processing engine is controlled by accuracy information of the position information.

Apparatus and method for generating an overview image of a plurality of images using an accuracy information

The paper presents an effective nonlinear adaptive switching non-local filter for the restoration of impulse-corrupted digital images by using distinct impulse detection and correction stages. The correction scheme of the filter adaptively switches between details-preserving non-local mode and signal restoration-based local mode to facilitate high fidelity in the restored image. The non-local filtering operation replaces impulses with a remote pixel that better suits the local image conditions. The algorithm works in this non-local mode only when there are sufficient uncorrupted pixels in the local neighborhood of the corrupted pixel to be replaced. Otherwise, the algorithm replaces impulsive pixels with the median of the uncorrupted pixels from the local neighborhood. Experimental results from various impulse noise levels support the improved performance of the proposed algorithm over other algorithms both subjectively and objectively.

Adaptive Switching Non-local Filter for the Restoration of Salt and Pepper Impulse-Corrupted Digital Images

In this paper we survey thoroughly the problem of orthorectified and incremental image mosaicking of a sequence of aerial images taken from low-altitude micro aerial vehicles. Most of existing approaches have been exploiting the global optimization (in presence of a loop in the image sequences) to distribute and/or metadata to mitigate the accumulating stitching error. However, the resulting mosaic can be improved if the errors are diminished by studying their sources. Mostly the UAV aerial image mosaicking is affected by the following three important sources of error: i) a weak homography as a result of using unleveled ground control points (GCPs) for image registration, ii) a poor camera calibration and image rectification, and iii) deficiency of a well-defined projection model (cylindrical, planar, etc) and consequently an inappropriate transformation model. We investigate the influences of using a depth map to find the features from the same plane, geometric distortion correction and combining the appropriate choice of projection and transformation model for the mosaicking. We further quantify the improvement of orthorectification in mosaics by mitigating those errors and demonstrate the improvement on real-world mosaics.

Incremental, orthorectified and loop-independent mosaicking of aerial images taken by micro UAVs

A mosaic method was put forward based on image wavelet transform and low-frequency area feature matching, which realized fast dynamic stitching of unmanned aerial vehicle's (UAV) sequence remote sensing images. This method made use of wavelet transform to obtain low-frequency image, searched and picked up character template image in this area, calculated and compared by using sequential similarity detection algorithm (SSDA), then obtained precise corresponding position relationship among the images. Based on the result of the image matching, the two images are stitched. Simulation experimental results show that the algorithm presented in the paper greatly improves the operation speed, while the precision remains fine, so it can be applied in real-time mosaicking of serial images from UAV very well.

A Mosaic Approach for Remote Sensing Images Based on Wavelet Transform

A new method of image denoising was put forward, based on integrated utilization of Curvelet transform and Cycle spinning. Because there existed "encircling" phenomenon for Curvelet transform, if images threshold denoising of Curvelet coefficients was performed directly, the image would be distorted, in this paper, Cycle spinning was introduced to eliminate the image distortion produced by "encircling". The simulation experiment indicated that the algorithm had perfect inhibiting ability for noise, moreover, it can protect the detail information of image; when different kinds of impulse noise images with different density were filtered, satisfied results can always be obtained.

Yuanhang Cheng

Papers

A Mosaic Approach for Remote Sensing Images Based on Wavelet Transform

Image Denoising Method Based on Curvelet Cycle Spinning