Edward Hsiao

Making specific features less discriminative to improve point-based 3D object recognition

We present a framework that retains ambiguity in feature matching to increase the performance of 3D object recognition systems. Whereas previous systems removed ambiguous correspondences during matching, we show that ambiguity should be resolved during hypothesis testing and not at the matching phase. To preserve ambiguity during matching, we vector quantize and match model features in a hierarchical manner. This matching technique allows our system to be more robust to the distribution of model descriptors in feature space. We also show that we can address recognition under arbitrary viewpoint by using our framework to facilitate matching of additional features extracted from affine transformed model images. The evaluation of our algorithms in 3D object recognition is demonstrated on a difficult dataset of 620 images.

Occlusion Reasoning for Object Detectionunder Arbitrary Viewpoint

We present a unified occlusion model for object instance detection under arbitrary viewpoint. Whereas previous approaches primarily modeled local coherency of occlusions or attempted to learn the structure of occlusions from data, we propose to explicitly model occlusions by reasoning about 3D interactions of objects. Our approach accurately represents occlusions under arbitrary viewpoint without requiring additional training data, which can often be difficult to obtain. We validate our model by incorporating occlusion reasoning with the state-of-the-art LINE2D and Gradient Network methods for object instance detection and demonstrate significant improvement in recognizing texture-less objects under severe occlusions.

Occlusion reasoning for object detection under arbitrary viewpoint

We present a unified occlusion model for object instance detection under arbitrary viewpoint. Whereas previous approaches primarily modeled local coherency of occlusions or attempted to learn the structure of occlusions from data, we propose to explicitly model occlusions by reasoning about 3D interactions of objects. Our approach accurately represents occlusions under arbitrary viewpoint without requiring additional training data, which can often be difficult to obtain. We validate our model by extending the state-of-the-art LINE2D method for object instance detection and demonstrate significant improvement in recognizing textureless objects under severe occlusions.

An optical tweezer actuated, nanoaperture-grid based Optofluidic Microscope implementation method

We report a novel grid based Optofluidic Microscope (OFM) method where a closely spaced 2D grid of nanoapertures (diameter = 100 nm, separation = 2.5 mum) provided patterned illumination. We achieved a one-to-one mapping of the light transmissions through the nanoapertures onto a high-speed CCD camera. By optically tweezing a targeted sample across the grid in a controlled fashion and recording the time varying light reception from the nanoapertures, we were able to generate high-resolution images of the sample. The achievable resolution limit of the prototype was ~ 110 nm (Sparrows criterion) under optimal conditions. We demonstrated the technique by imaging polystyrene beads and pollen spores.

Car make and model recognition using 3D curve alignment

We present a new approach for recognizing the make and model of a car from a single image. While most previous methods are restricted to fixed or limited viewpoints, our system is able to verify a cars make and model from an arbitrary view. Our model consists of 3D space curves obtained by backprojecting image curves onto silhouette-based visual hulls and then refining them using three-view curve matching. These 3D curves are then matched to 2D image curves using a 3D view-based alignment technique. We present two different methods for estimating the pose of a car, which we then use to initialize the 3D curve matching. Our approach is able to verify the exact make and model of a car over a wide range of viewpoints in cluttered scenes.

Shape-Based Instance Detection Under Arbitrary Viewpoint

Shape-based instance detection under arbitrary viewpoint is a very challenging problem. Current approaches for handling viewpoint variation can be divided into two main categories: invariant and non-invariant. Invariant approaches explicitly represent the structural relationships of high-level, view-invariant shape primitives. Non-invariant approaches, on the other hand, create a template for each viewpoint of the object, and can operate directly on low-level features. We summarize the main advantages and disadvantages of invariant and non-invariant approaches, and conclude that non-invariant approaches are well-suited for capturing fine-grained details needed for specific object recognition while also being computationally efficient. Finally, we discuss approaches that are needed to address ambiguities introduced by recognizing shape under arbitrary viewpoint.

A high resolution OptoFluidic microscope with optical tweezer actuation

In this paper, we will report our recent development of a new type of OptoFluidic Microscope (OFM) that is capable of delivering resolution beyond the diffraction limit of light. Accurate control of the sample translation is accomplished by adopting an optical tweezer scanner into the system. During the image acquisition, a two-dimensional nanoaperture array defined on a thin aluminum film acts as an array of ultra-fine illumination sources. The imaging system is tested and demonstrated by using polystyrene beads and green algae (Chlamydomonas). Properties of the system are reported and discussed.

Coherent Occlusion Reasoning for Instance Recognition

Occlusions are common in real world scenes and are a major obstacle to robust object detection. In this paper, we present a method to coherently reason about occlusions on many types of detectors. Previous approaches primarily enforced local coherency or learned the occlusion structure from data. However, local coherency ignores the occlusion structure in real world scenes and learning from data requires tediously labeling many examples of occlusions for every view of every object. Other approaches require binary classifications of matching scores. We address these limitations by formulating occlusion reasoning as an efficient search over occluding blocks which best explain a probabilistic matching pattern. Our method demonstrates significant improvement in estimating the mask of the occluding region and improves object instance detection on a challenging dataset of objects under severe occlusions.