Search

With the advent of 3D cameras, getting depth information along with RGB images has been facilitated, which is helpful in various computer vision tasks. However, there are two challenges in using these RGB-D images to help recognize RGB images captured by conventional cameras: one is that the depth images are missing at the testing stage, the other is that the training and test data are drawn from different distributions as they are captured using different equipment. To jointly address the two challenges, we propose an asymmetrical transfer learning framework, wherein three classifiers are trained using the RGB and depth images in the source domain and RGB images in the target domain with a structural risk minimization criterion and regularization theory. A cross-modality co-regularizer is used to restrict the two-source classifier in a consistent manner to increase accuracy. Moreover, an L2,1 norm cross-domain co-regularizer is used to magnify significant visual features and inhibit insignificant ones in the weight vectors of the two RGB classifiers. Thus, using the cross-modality and cross-domain co-regularizer, the knowledge of RGB-D images in the source domain is transferred to the target domain to improve the target classifier. The results of the experiment show that the proposed method is one of the most effective ones.

Reinforcement learning (RL) constitutes an effective method of controlling dynamic systems without prior knowledge. One of the most important and difficult problems in RL is the improvement of data efficiency. Probabilistic inference for learning control (PILCO) is a state-of-the-art data-efficient framework that uses a Gaussian process to model dynamic systems. However, it only focuses on optimizing cumulative rewards and does not consider the accuracy of a dynamic model, which is an important factor for controller learning. To further improve the data efficiency of PILCO, we propose its active exploration version (AEPILCO) that utilizes information entropy to describe samples. In the policy evaluation stage, we incorporate an information entropy criterion into long-term sample prediction. Through the informative policy evaluation function, our algorithm obtains informative policy parameters in the policy improvement stage. Using the policy parameters in the actual execution produces an informative sample set; this is helpful in learning an accurate dynamic model. Thus, the AEPILCO algorithm improves data efficiency by learning an accurate dynamic model by actively selecting informative samples based on the information entropy criterion. We demonstrate the validity and efficiency of the proposed algorithm for several challenging controller problems involving a cart pole, a pendubot, a double pendulum, and a cart double pendulum. The AEPILCO algorithm can learn a controller using fewer trials compared to PILCO. This is verified through theoretical analysis and experimental results.

Image segmentation is an important processing in many applications such as image retrieval and computer vision. One of the most successful models for image segmentation is the level set methods which are based on local context. The methods, though comparatively effective in segmenting images with inhomogeneous intensity, are considerably computation-intensive and at the risk of falling into local minima in the convergence of the active contour energy function. To address the issues, we propose a region-based level set method, called KL-MLBF, which is based on the multi-scale local binary fitting (MLBF) and the Kullback-Leibler (KL) divergence. We first apply the multi-scale theory to the local binary fitting model to build MLBF. Then the energy term measured by KL divergence between regions to be segmented is incorporated into the energy function of MLBF. KL-MLBF utilizes the between-cluster distance and the adaptive kernel function selection strategy to formulate the energy function. Being more robust to the initial location of the contour than the classical segmentation models, KL-MLBF can deal with blurry boundaries and noise problems. The results of experiments on synthetic and real images have shown that KL-MLBF can improve the effectiveness of segmentation while ensuring the accuracy by accelerating the minimization of the energy function. [ABSTRACT FROM AUTHOR]

A fast line segment detection method, unite-and-divide (UND) approach, is investigated, which includes two phases, namely the union of spectra in the frequency domain and the division of the sinogram in Radon space respectively. In the union phase, for a given image, its sinogram is obtained by parallel 2D multilayer Fourier transform, Cartesian-to-polar mapping and ID inverse Fourier transform. In the division phase, every peaks and edges of butterfly wings in its neighborhood in sinogram are firstly specified, with each neighborhood area corresponding to a window in image space, and then, by applying edge-analyzing to each sinogram of each windowed separately, the endpoints of line segments are extracted. Our experiments are conducted on benchmark images, the results reveal that the UND method yields high accuracy, low computational cost and is more robust to noise. [ABSTRACT FROM AUTHOR]