尺度空间检测边缘特征的进化分析

多尺度分析中的一个重要问题是检测特征的尺度空间进化规律。本文以2-D高斯函数作为核函数,明确给出了角点图像的尺度空间检测边缘是一个由角边界的斜率和分析尺度共同决定的椭圆。以此为基础,讨论了尺度因子对相邻角点检测边缘的同化效应,给出了角最小分辨率与分析尺度的关系不等式。本文的结果对多尺度分析具有普遍性参考意义。     

Landmark-Based Registration of Curves via the Continuous Wavelet Transform

This article is concerned with the problem of the alignment of multiple sets of curves. We analyze two real examples arising from the biomedical area for which we need to test whether there are any statistically significant differences between two subsets of subjects. To synchronize a set of curves, we propose a new nonparametric landmark-based registration method based on the alignment of the structural intensity of the zero-crossings of a wavelet transform. The structural intensity is a recently proposed multiscale technique that highlights the main features of a signal observed with noise. We conduct a simulation study to compare our landmark-based registration approach with some existing methods for curve alignment. For the two real examples, we compare the registered curves with FANOVA techniques, and a detailed analysis of the warping functions is provided.     

A panorama on multiscale geometric representations, intertwining spatial, directional and frequency selectivity

The richness of natural images makes the quest for optimal representations in image processing and computer vision challenging. The latter observation has not prevented the design of image representations, which trade off between efficiency and complexity, while achieving accurate rendering of smooth regions as well as reproducing faithful contours and textures. The most recent ones, proposed in the past decade, share a hybrid heritage highlighting the multiscale and oriented nature of edges and patterns in images. This paper presents a panorama of the aforementioned literature on decompositions in multiscale, multi-orientation bases or dictionaries. They typically exhibit redundancy to improve sparsity in the transformed domain and sometimes its invariance with respect to simple geometric deformations (translation, rotation). Oriented multiscale dictionaries extend traditional wavelet processing and may offer rotation invariance. Highly redundant dictionaries require specific algorithms to simplify the search for an efficient (sparse) representation. We also discuss the extension of multiscale geometric decompositions to non-Euclidean domains such as the sphere or arbitrary meshed surfaces. The etymology of panorama suggests an overview, based on a choice of partially overlapping “pictures”. We hope that this paper will contribute to the appreciation and apprehension of a stream of current research directions in image understanding.     

Small Target Detection Utilizing Robust Methods of the Human Visual System for IRST

Robust detection of small targets is very important in IRST (Infrared Search and Track). This paper presents a novel mathematical method for the incoming target detection problem in cluttered background motivated from the robust properties of human visual system (HVS). The HVS shows the best efficiency and robustness for an object detection task. The robust properties of the HVS are contrast mechanism, multi-resolution representation, size adaptation, and pop-out phenomena. Based on these facts, a plausible computational model integrating these facts is proposed using Laplacian scale-space theory and Tune-Max based optimization method. Simultaneous target signal enhancement and background clutter suppression is achieved by tuning and maximizing the signal-to-clutter ratio (TMSCR) in Laplacian scale-space. At the first stage, the Tune-Max of the signal to background contrast produces candidate targets with adapted scale. At the second stage, the Tune-Max of the signal-to-clutter ratio (SCR) produces maximal SCR which is used to pop-out detections. Experimental evaluation results for the incoming target sequence validate the upgraded detection capability of the proposed method compared with the Top-hat method at the same false alarm rate. Experimental results for the six kinds of cluttered background images show that the proposed TMSCR produces less false alarms (4.3 times reduction) compared to the Top-hat at the same detection rate.