Maximal covering, Backup covering, Optimization, Geographic information systems, Police patrol areas, Police beats,

We developed a neural network based approach to identify urban tree species at the individual tree level from lidar and hyperspectral imagery. This approach is capable of modeling the characteristics of multiple spectral signatures within each species using an internally unsupervised engine, and is able to catch spectral differences between species using an externally supervised system. To generate a species-level map for an urban forest with high spatial heterogeneity and species diversity, we conducted a treetop-based species identification. This can avoid the problems of double-sided illumination, shadow, and mixed pixels, encountered in the crown-based species classification. The study indicates lidar data in conjunction with hyperspectral imagery are not only capable of detecting individual trees and estimating their tree metrics, but also identifying their species types using the developed algorithm. The integration of these two data sources has great potential to take the place of traditional field surveys.

Abstract:The objective of this study is to develop new algorithms for automated urban forest inventory at the individual tree level using LiDAR point cloud data. LiDAR data contain three-dimensional structure information that can be used to estimate tree height, base height, crown depth, and crown diameter. This allows precision urban forest inventory down to individual trees. Unlike most of the published algorithms that detect individual trees from a LiDAR-derived raster surface, we worked directly with the LiDAR point cloud data to separate individual trees and estimate tree metrics. Testing results in typical urban forests are encouraging. Future works will be oriented to synergize LiDAR data and optical imagery for urban tree characterization through data fusion techniques.

Neural networks, which make no assumption about data distribution, have achieved improved image classification results compared to traditional methods. Unfortunately, a neural network is generally perceived as being a 'black box'. It is extremely difficult to document how specific classification decisions are reached. Fuzzy systems, on the other hand, have the capability to represent classification decisions explicitly in the form of fuzzy 'if-then' rules. However, the construction of a knowledge base, especially the fine-tuning of the fuzzy set parameters of the fuzzy rules in a fuzzy expert system, is a tedious and subjective process. This research has developed a new, improved neuro-fuzzy image classification system based on the synergism between neural networks and fuzzy expert systems. It incorporates the best of both technologies and compensates for the shortcomings of each. The learning algorithms of neural networks developed here are used to automate the derivation of fuzzy set parameters for the fuzzy 'if-then' rules in a fuzzy expert system. The rules obtained, in symbolic form, facilitate the understanding of the neural network based image classification system. In addition, the image classification accuracy obtained from the improved neuro-fuzzy system was significantly superior to those of the back-propagation based neural network and the maximum likelihood approaches. IntroductionData acquired by remote sensing systems can be used to systematically extract a variety of fundamental biophysical and land use/land cover information over large geographic areas if processed with the appropriate analytical tools (Jensen 2000). The recent availability of high spatial resolution imagery (e.g. Space Imaging Inc. IKONOS 1m61 m panchromatic and 4 m64 m multispectral) and hyperspectral data (e.g. NASA's Hyperion sensor system) may be very useful for extracting land use and land cover information (Jensen and Cowen 1999). However, the higher spatial resolution imagery now contains even more high frequency information. Similarly, the greater number of bands introduces additional redundancy in information content and vastly complicates the task of image processing. In effect,

No abstract