INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING quarterly report for Application of Forest Image Analysis to Monitoring and Modeling of Psychological, Silvicultural, and Wildlife Habitat Attributes

This quarter we primarily worked on optimizing the edge and line detectors incorporated in our image analysis software. We changed our optimization scenario and designed a metric for the evaluation procedure. With our optimized detectors, we achieved a match error rate of 29.3%, as well as a low level of insertion, on the test set 01 from the Pre-phase 01 image data. Meanwhile, we started investigating the frequency analysis technique of Gabor filters. We are now in the process of replicating a system which uses Gabor filters to do unsupervised texture segmentation, in an attempt to understand the behavior of Gabor filters.