Terry Fong

Vehicle detection from aerial imagery

Vehicle detection from aerial images is becoming an increasingly important research topic in surveillance, traffic monitoring and military applications. The system described in this paper focuses on vehicle detection in rural environments and its applications to oil and gas pipeline threat detection. Automatic vehicle detection by unmanned aerial vehicles (UAV) will replace current pipeline patrol services that rely on pilot visual inspection of the pipeline from low altitude high risk flights that are often restricted by weather conditions. Our research compares a set of feature extraction methods applied for this specific task and four classification techniques. The best system achieves an average 85% vehicle detection rate and 1800 false alarms per flight hour over a large variety of areas including vegetation, rural roads and buildings, lakes and rivers collected during several day time illuminations and seasonal changes over one year

An Edge-Less Approach to Horizon Line Detection

Horizon line is a promising visual cue which can be exploited for robot localization or visual geo-localization. Prominent approaches to horizon line detection rely on edge detection as a pre-processing step which is inherently a non-stable approach due to parameter choices and underlying assumptions. We present a novel horizon line detection approach which uses machine learning and Dynamic Programming (DP) to extract the horizon line from a classification map instead of an edge map. The key idea is assigning a classification score to each pixel, which can be interpreted as the likelihood of the pixel belonging to the horizon line, and representing the classification map as a multi-stage graph. Using DP, the horizon line can be extracted by finding the path that maximizes the sum of classification scores. In contrast to edge maps which are typically binary (edge vs no-edge) and contain gaps, classification maps are continuous and contain no gaps, yielding significantly better solutions. Using classification maps instead of edge maps allows for removing certain assumptions such as the horizon is close to the top of the image or that the horizon forms a straight line. The purpose of these assumptions is to bias the DP solution but they fail to produce good results when they are not valid. We demonstrate our approach on three different data sets and provide comparisons with a traditional approach based on edge maps. Although our training set is comprised of a very small number of images from the same location, our results illustrate that our method generalizes well to images acquired under different conditions and geographical locations.

Coupling Dynamic Programming with Machine Learning for Horizon Line Detection

In this paper, we consider the problem of segmenting an image into sky and non-sky regions, typically referred to as horizon line detection or skyline extraction. Specifically, we present a new approach to horizon line detection by coupling machine learning with dynamic programming. Given an image, the Canny edge detector is applied first and keeping only those edges which survive over a wide range of thresholds. We refer to the surviving edges as Maximally Stable Extremal Edges (MSEEs). The number of edges is further reduced by classifying MSEEs into horizon and non-horizon edges using a Support Vector Machine (SVM) classifier. Dynamic programming is then applied on the horizon classified edges to extract the horizon line. Various local texture features and their combinations have been investigated in training the horizon edge classifier including SIFT, LBP, HOG, SIFT-LBP, SIFT-HOG, LBP-HOG and SIFT-LBP-HOG. We have also investigated various nodal costs in the context of dynamic programming including binary edge scores, normalized edge classification scores, gradient magnitude and their combinations. The proposed approach has been evaluated and compared with a competitive approach on two challenging data sets, illustrating superior performance.

An Experimental Evaluation of Different Features and Nodal Costs for Horizon Line Detection

Horizon line detection is a segmentation problem where a boundary between a sky and non-sky region is searched. Conventionally edge detection is performed as the first step followed by dynamic programming to find the shortest path which conforms to the detected horizon line. Recent work has proposed the use of machine learning to reduce the number of non-horizon edges to accurately detect the horizon line. In this paper, we investigate the suitablity of various local texture features and their combinations to reduce the number of false classifications for a recently proposed horizon detection approach. Specifically, we explore SIFT, LBP, HOG and their combinations SIFT-LBP, SIFT-HOG, LBP-HOG and SIFT-LBP-HOG as features to train the SVM classifier. We further show that using only edge information as the nodal costs is not enough and propose various nodal costs which can result in enhanced accuracy of the detected horizon line as evidenced by the conducted experiments and results. We compare our proposed formulations with an earlier approach relying only on edges and suffers due to faulty assumptions. We report our comparative results for an image set comprising of mountainous images captured during an outdoor robot exploration of Basalt Hills.

Automatic Crater Detection Using Convex Grouping and Convolutional Neural Networks

Craters are some the most important landmarks on the surface of many planets which can be used for autonomous safe landing and spacecraft and rover navigation. Manual detection of craters is laborious and impractical, and many approaches have been proposed in the field to automate this task. However, none of these methods have yet become a standard tool for crater detection due to the challenging nature of this problem. In this paper, we propose a new crater detection algorithm (CDA) which employs a multi-scale candidate region detection step based on convexity cues and candidate region verification based on machine learning. Using an extensive dataset, our method has achieved a 92 % detection rate with an 85 % precision rate.

On Crater Verification Using Mislocalized Crater Regions

Automatic crater detection in planetary images is an important task with many applications in planetary science, spacecraft navigation, landing, and control. Typically, crater detection algorithms consist of two main steps: candidate crater region extraction and crater verification. Various methods have been proposed for extracting candidate crater regions, ranging from detecting circular/elliptical regions to detecting highlight and shadow regions. For crater verification, powerful feature extraction and machine learning techniques have been employed. While this two-step approach can be efficient and robust, inaccuracies in the candidate crater region extraction step can result in mislocalized crater regions which could affect verification performance. In this paper, we investigate the robustness of various feature extraction methods to mislocalized crater regions. Using features which are robust to localization errors but also choosing a more representative training set has yielded significant performance improvements on an extensive dataset from the Lunar Reconnaissance Orbiter (LRO).

Human Exploration of Asteroids, the Moon, and Mars Using Robotic Arm-Equipped Pressurized Vehicles

During the 2011 summer field campaign of the NASA Haughton-Mars Project at the Haughton impact crater site on Devon Island in the high Arctic, field tests were conducted of the use of a robotic arm system integrated to Humvees serving as analog pressurized vehicles for the human exploration of near-Earth asteroids, the Moon, and Mars. The goal of these field tests was to begin assessing how field geology, including sample acquisition, might in some cases be conducted effectively from the confines of a pressurized vehicle (in intra-vehicular or IVA mode) without the crew having to always go out on EVA. Preliminary findings suggest that, particularly at sites where rocks and soils available for sampling occur mostly as loose rubble or float (as with planetary regoliths), the use of a robotic arm can allow efficient collection and characterization of high quality samples. An important implication of the finding for future space exploration architecture development, if confirmed through further field tests, is that the number and frequency of EVAs needed for sample acquisition during planetary vehicular traverses might be reduced from current assumptions. Wear and tear on EVA suit and suit-port systems, mission operations complexity, crew exposure to space radiation, and planetary protection concerns, might all be significantly reduced if robotic-arm assisted sample acquisition in IVA mode became available as an effective option for science operations.

Lunar terrain and albedo reconstruction of the apollo 15 zone

Generating accurate three dimensional planetary models is becoming increasingly important as NASA plans manned missions to return to the Moon in the next decade. This paper describes a 3D surface and albedo reconstruction from orbital imagery. The techniques described here allow us to automatically produce seamless, highly accurate digital elevation and albedo models from multiple stereo image pairs while significantly reducing the influence of image noise. Our technique is demonstrated on the entire set of orbital images retrieved by the Apollo 15 mission.