Crater detection from commercial satellite imagery to estimate unexploded ordnance in Cambodian agricultural land

Abstract

Unexploded ordnance (UXO) pose a significant threat to post-conflict communities, and current efforts to locate bombs rely on time-intensive and dangerous in-person enumeration. Very high resolution (VHR) sub-meter satellite images may offer a low-cost and high-efficiency approach to automatically detect craters and estimate UXO density. Machine-learning methods from the meteor crater literature are ill-suited to find bomb craters, which are smaller than meteor craters and have high appearance variation, particularly in spectral reflectance and shape, due to the complex terrain environment. A two-stage learning-based framework is created to address these challenges. First, a simple and loose statistical classifier based on histogram of oriented gradient (HOG) and spectral information is used for a first pass of crater recognition. In a second stage, a patch-dependent novel spatial feature is developed through dynamic mean-shift segmentation and SIFT descriptors. We apply the model to a multispectral WorldView-2 image of a Cambodian village, which was heavily bombed during the Vietnam War. The proposed method increased true bomb crater detection by over 160 percent. Comparative analysis demonstrates that our method significantly outperforms typical object-recognition algorithms and can be used for wide-area bomb crater detection. Our model, combined with declassified records and demining reports, suggests that 44 to 50 percent of the bombs in the vicinity of this particular Cambodian village may remain unexploded.

Aerial Bombing Crater Identification: Exploitation of Precise Digital Terrain Models

“…In other words, the identification of terrestrial craters resulted in a large number of false positives. It is highly probable that the poor results can be attributed to differences in the morphology and size of the extraterrestrial craters [14]. Although the tool is trained on lunar craters with generally larger (up to subkilometer) sizes and sharp rims, terrestrial craters, even in the case of highly precise LiDAR-derived DTM, usually do not display such properties.…”

“…This method is regularly used for land use and land cover mapping [11], but a growing number of studies present applications for smaller object detection [12]. Implementation of the GEOBIA approach was tested in the case of detection of charcoal hearths [13], aerial bombing crater detection based on precise satellite imagery in agricultural fields [14], and detection of WWI shell craters [15] based on LiDAR-derived DTM and sky-view-factor hill shading. Regardless of the feature under research, the initial or later phases of identification deal with the edge detection technique for the sake of the division of the object from its surroundings.…”

“…Regardless of the feature under research, the initial or later phases of identification deal with the edge detection technique for the sake of the division of the object from its surroundings. Several filter types are available for edge recognition and have been applied to the recognition of ancient linear features (Gabor filters, [16]) or feature shape-independent histograms of gradients with applications to craters [14].…”

“…Despite the similar morphology of the objects in the study (circular impact craters), the use of automated tools for terrestrial bombing crater detection is not perceived as viable. The factors that can disqualify the tool replicability for terrestrial bombing craters are usually based on the presence of vegetation, erosion of the bombing crater edges, and different sizes of the bombing craters [14]. Such factors lead, inter alia, to testing and development of tools directly on (terrestrial) bombing craters under research (e.g., [26]).…”

Virome in adult Aedes albopictus captured during different seasons in Guangzhou City, China

“…Outbreaks caused by arboviruses have been reported, such as dengue fever in Thailand, Vietnam, and China, and chikungunya in France, Tanzania, Africa, and the Americas [ 2 , 6 – 13 ]. ZIKV infections have also been reported in many countries, including Brazil, Nicaragua, and America [ 14 – 16 ].…”