We present in this paper a novel study aiming at identifying the differences in visual search patterns between physicians of diverse levels of expertise during the screening of colonoscopy videos. Physicians were clustered into two groups -experts and novices- according to the number of procedures performed, and fixations were captured by an eye-tracker device during the task of polyp search in different video sequences. These fixations were integrated into heat maps, one for each cluster. The obtained maps were validated over a ground truth consisting of a mask of the polyp, and the comparison between experts and novices was performed by using metrics such as reaction time, dwelling time and energy concentration ratio. Experimental results show a statistically significant difference between experts and novices, and the obtained maps show to be a useful tool for the characterisation of the behaviour of each group.

We present in this paper a novel study aiming at identifying the differences in visual search patterns between physicians of diverse levels of expertise during the screening of colonoscopy videos. Physicians were clustered into two groups -experts and novices- according to the number of procedures performed, and fixations were captured by an eye-tracker device during the task of polyp search in different video sequences. These fixations were integrated into heat maps, one for each cluster. The obtained maps were validated over a ground truth consisting of a mask of the polyp, and the comparison between experts and novices was performed by using metrics such as reaction time, dwelling time and energy concentration ratio. Experimental results show a statistically significant difference between experts and novices, and the obtained maps show to be a useful tool for the characterisation of the behaviour of each group.

The estimation of covariance matrices is a crucial step in several statistical tasks. Especially when using few samples of a high dimensional representation of shapes, the standard maximum likelihood estimation (ML) of the covariance matrix can be far from the truth, is often rank deficient, and may lead to unreliable results. In this paper, we discuss regularization by prior knowledge using maximum a posteriori (MAP) estimates. We compare ML to MAP using a number of priors and to Tikhonov regularization. We evaluate the covariance estimates on both synthetic and real data, and we analyze the estimates' influence on a missing-data reconstruction task, where high resolution vertebra and cartilage models are reconstructed from incomplete and lower dimensional representations. Our results demonstrate that our methods outperform the traditional ML method and Tikhonov regularization.

A variety of modifications have been employed to learning vector quantization (LVQ) algorithms using either crisp or soft windows for selection of data. Although these schemes have been shown in practice to improve performance, a theoretical study on the influence of windows has so far been limited. Here we rigorously analyze the influence of windows in a controlled environment of gaussian mixtures in high dimensions. Concepts from statistical physics and the theory of online learning allow an exact description of the training dynamics, yielding typical learning curves, convergence properties, and achievable generalization abilities. We compare the performance and demonstrate the advantages of various algorithms, including LVQ 2.1, generalized LVQ (GLVQ), Learning from Mistakes (LFM) and Robust Soft LVQ (RSLVQ). We find that the selection of the window parameter highly influences the learning curves but not, surprisingly, the asymptotic performances of LVQ 2.1 and RSLVQ. Although the prototypes of LVQ 2.1 exhibit divergent behavior, the resulting decision boundary coincides with the optimal decision boundary, thus yielding optimal generalization ability.

Psychophysical researches on the human visual system have shown that the points of high curvature on the contour of an object play an important role in the recognition process. Inspired by these studies we propose: (i) a novel algorithm to select points of high curvature on the contour of an object which can be used to construct a recognizable polygonal approximation, (ii) a test which evaluates the effect of deletion of contour segments containing such points on the performance of contour based object recognition algorithms. We use complete contour representations of objects as a reference (training) set. Incomplete contour representations of the same objects are used as a test set. The performance of an algorithm is reported using the recognition rate as a function of the percentage of contour retained. We consider two types of contour incompleteness obtained by deletion of contour segments of high or low curvature. We illustrate the test procedure using two shape recognition algorithms that deploy a shape context and a distance multiset as local shape descriptors. Both algorithms qualitatively mimic human visual perception in that the deletion of segments of high curvature has a stronger performance degradation effect than the deletion of other parts of the contour. This effect is more pronounced in the performance of the shape context method.

Learning vector quantization (LVQ) constitutes a powerful and intuitive method for adaptive nearest prototype classification. However, original LVQ has been introduced based on heuristics and numerous modifications exist to achieve better convergence and stability. Recently, a mathematical foundation by means of a cost function has been proposed which, as a limiting case, yields a learning rule similar to classical LVQ2.1. It also motivates a modification which shows better stability. However, the exact dynamics as well as the generalization ability of many LVQ algorithms have not been thoroughly investigated so far. Using concepts from statistical physics and the theory of on-line learning, we present a mathematical framework to analyse the performance of different LVQ algorithms in a typical scenario in terms of their dynamics, sensitivity to initial conditions, and generalization ability. Significant differences in the algorithmic stability and generalization ability can be found already for slightly different variants of LVQ. We study five LVQ algorithms in detail: Kohonen's original LVQ1, unsupervised vector quantization (VQ), a mixture of VQ and LVQ, LVQ2.1, and a variant of LVQ which is based on a cost function. Surprisingly, basic LVQ1 shows very good performance in terms of stability, asymptotic generalization ability, and robustness to initializations and model parameters which, in many cases, is superior to recent alternative proposals.

With inspiration from psychophysical researches of the human visual system, we propose a novel aspect and a method for performance evaluation of contour-based shape recognition algorithms regarding their robustness to incompleteness of contours. We use complete contour representations of objects as a reference (training) set. Incomplete contour representations of the same objects are used as a test set. The performance of an algorithm is reported using the recognition rate as a function of the percentage of contour retained. We call this evaluation procedure the ICR test. We consider three types of contour incompleteness, viz. segment-wise contour deletion, occlusion, and random pixel depletion. As an illustration, the robustness of two shape recognition algorithms to contour incompleteness is evaluated. These algorithms use a shape context and a distance multiset as local shape descriptors. Qualitatively, both algorithms mimic human visual perception in the sense that recognition performance monotonously increases with the degree of completeness and that they perform best in the case of random depletion and worst in the case of occluded contours. The distance multiset method performs better than the shape context method in this test framework.

With inspiration from psychophysical researches of the human visual system we propose a novel method for performance evaluation of contour based shape recognition algorithms. We use complete contour representations of objects as a training set. Incomplete contour representations of the same objects are used as a test set and the recognition performance of two shape based methods is investigated. The amount of incompleteness in test cases is quantified using the percentage of contour pixels retained. The performances of the methods are reported using the recognition rate as a function of the degree of incompleteness. We consider three types of incomplete contour representations, viz. segment-wise deletion, occlusion and random pixel depletion. The methods compared in this framework use shape context and distance multiset as local shape descriptors. Qualitatively, both methods mimic human visual perception in the sense that they perform best in the case of random depletion and worst in the case of occluded contours. Quantitatively, the distance multiset method performs better than the shape context method in this test framework.