The moving object detection with moving camera has the problems that it is difficult to model the background and the computation cost is usually high. In order to solve the problems, a method for detecting moving object with moving camera based on motion saliency was proposed, which realized accurate moving object detection and avoided complex background modeling. The moving objects were detected according to the saliency of the video scene, which was computed based on the simulation of the attention mechanism in human vision system and the moving properties of background and foreground when the camera moved in translation. Firstly, the motion features of object were extracted by optical flow method and the background motion texture was suppressed by 2-D Gaussian convolution. Then the global saliency of motion features was measured by counting the histogram. According to the temporal salient map, the color information of foreground and background was extracted respectively. Finally, Bayesian model was used to deal with temporal salient map for extracting salient moving objects. The experimental results on the public video datasets show that the proposed method can suppress background motion noise, while detecting the moving object distinctly and accurately in the dynamic scene with moving camera.

Concerning the difficulty of its similarity to the expression and the effects of noise in medical image retrieval, a diffusion-based approach on a tensor product graph was proposed to improve the texton-based pairwise similarity metric by context information of other database objects. Firstly, medical image features were described and extracted by texton-based statistical method, and then the pairwise similarities were obtained with weights determined by the similarities between textons. A global similarity metric was achieved by utilizing the tensor product graph to propagate the similarity information along the intrinsic structure of the data manifold. Experimental results of ImageCLEFmed 2009 database show that, the proposed algorithm improves the performance by an average class accuracy of 32% and 19% compared with the Gabor-based retrieval algorithm and the Scale-Invariant Feature Transform (SIFT)-based retrieval algorithm respectively, which can be applied to medical image retrieval.