Representation For Object Recognition Research Articles

Psychophysical evaluation of planar shape representations for object recognition

Psychophysical evaluation of planar shape representations for object recognition

Efficient image representation for object recognition via pivots selection

Patch-level features are essential for achieving good performance in computer vision tasks. Besides wellknown pre-defined patch-level descriptors such as scaleinvariant feature transform (SIFT) and histogram of oriented gradient (HOG), the kernel descriptor (KD) method [1] offers a new way to "grow-up" features from a match-kernel defined over image patch pairs using kernel principal component analysis (KPCA) and yields impressive results. In this paper, we present efficient kernel descriptor (EKD) and efficient hierarchical kernel descriptor (EHKD), which are built upon incomplete Cholesky decomposition. EKD automatically selects a small number of pivot features for generating patch-level features to achieve better computational efficiency. EHKD recursively applies EKD to form image-level features layer-by-layer. Perhaps due to parsimony, we find surprisingly that the EKD and EHKD approaches achieved competitive results on several public datasets compared with other state-of-the-art methods, at an improved efficiency over KD.

Categorizing Dynamic Textures Using a Bag of Dynamical Systems

We consider the problem of categorizing video sequences of dynamic textures, i.e., nonrigid dynamical objects such as fire, water, steam, flags, etc. This problem is extremely challenging because the shape and appearance of a dynamic texture continuously change as a function of time. State-of-the-art dynamic texture categorization methods have been successful at classifying videos taken from the same viewpoint and scale by using a Linear Dynamical System (LDS) to model each video, and using distances or kernels in the space of LDSs to classify the videos. However, these methods perform poorly when the video sequences are taken under a different viewpoint or scale. In this paper, we propose a novel dynamic texture categorization framework that can handle such changes. We model each video sequence with a collection of LDSs, each one describing a small spatiotemporal patch extracted from the video. This Bag-of-Systems (BoS) representation is analogous to the Bag-of-Features (BoF) representation for object recognition, except that we use LDSs as feature descriptors. This choice poses several technical challenges in adopting the traditional BoF approach. Most notably, the space of LDSs is not euclidean; hence, novel methods for clustering LDSs and computing codewords of LDSs need to be developed. We propose a framework that makes use of nonlinear dimensionality reduction and clustering techniques combined with the Martin distance for LDSs to tackle these issues. Our experiments compare the proposed BoS approach to existing dynamic texture categorization methods and show that it can be used for recognizing dynamic textures in challenging scenarios which could not be handled by existing methods.

BRIDGING STRUCTURE AND FEATURE REPRESENTATIONS IN GRAPH MATCHING

Structures and features are opposite approaches in building representations for object recognition. Bridging the two is an essential problem in pattern recognition as the two opposite types of information are fundamentally different. As dissimilarities can be computed for both the dissimilarity representation can be used to combine the two. Attributed graphs contain structural as well as feature-based information. Neglecting the attributes yields a pure structural description. Isolating the features and neglecting the structure represents objects by a bag of features. In this paper we will show that weighted combinations of dissimilarities may perform better than these two extremes, indicating that these two types of information are essentially different and strengthen each other. In addition we present two more advanced integrations than weighted combining and show that these may improve the classification performances even further.

Similarity dependency of the change in ERP component N1 accompanying with the object recognition learning

Performance during object recognition across views is largely dependent on inter-object similarity. The present study was designed to investigate the similarity dependency of object recognition learning on the changes in ERP component N1. Human subjects were asked to train themselves to recognize novel objects with different inter-object similarity by performing object recognition tasks. During the tasks, images of an object had to be discriminated from the images of other objects irrespective of the viewpoint. When objects had a high inter-object similarity, the ERP component, N1 exhibited a significant increase in both the amplitude and the latency variation across objects during the object recognition learning process, and the N1 amplitude and latency variation across the views of the same objects decreased significantly. In contrast, no significant changes were found during the learning process when using objects with low inter-object similarity. The present findings demonstrate that the changes in the variation of N1 that accompany the object recognition learning process are dependent upon the inter-object similarity and imply that there is a difference in the neuronal representation for object recognition when using objects with high and low inter-object similarity.

A Modified Exoskeleton for 3D Shape Description and Recognition.

Three-dimensional(3D) shape representation is a powerful tool in object recognition that is an essential process in an image processing and analysis system. Skeleton is one of the most widely used representations for object recognition, nevertheless most of the skeletons obtained from conventional methods are susceptible to rotation and noise disturbances. In this paper, we present a new 3D object representation called a modified exoskeleton (mES) which preserves skeleton properties including significant characteristics about an object that are meaningful for object recognition, and is more stable and less susceptible to rotation and noise than the skeletons. Then a 3D shape recognition methodology which determines the similarity between an observed object and other known objects in a database is introduced. Through a number of experiments on 3D artificial objects and real volumetric lung tumors extracted from CT images, it can be verified that our proposed methodology based on the mES is a simple yet efficient method that is less sensitive to rotation, noise, and independent of orientation and size of the objects.

An optimum solution for scale-invariant object recognition based on the multiresolution approximation

This paper presents a multiresolution approximation approach to obtaining boundary representations for object recognition. Our technique combines a multiresolution approximation and the curvature scale-space representation for obtaining representations. Our research consists of two main parts. In the first part of our research, we introduce the continuous multiresolution approximation (CMA) in terms of the continuous wavelet transform (CWT). Then we implement a fast algorithm to compute the CMA. We apply the CMA to a boundary to obtain approximations of the boundary at various resolutions. The CMA provides a consistent interpretation of objects with scale-variations. Moreover, we can quickly compute our representations by using the fast algorithm for the CMA. In the second part, we propose three representations for object recognition which cover most boundary-based object recognition problems. All three representations use the approximations obtained by the CMA. Each representation has different features and covers different types of matching problems but all representations are constructed by using curvature zero crossings of the approximations. Our representations provide a general but reliable solution to most boundary based object matching problems. Finally, we investigate the properties of our representations such as validity, efficiency, and reliability. We verified our results experimentally to demonstrate the feasibility of using our representations for object recognition.