传统的基于形状信息目标定位的算法,对目标观测角度发生形变情况下的定位存在不少困难, 针对该问题, 提出了一种基于稀疏活动轮廓模型的感兴趣目标(OOI)检测算法。首先通过共同勾画算法学习到感兴趣目标的稀疏活动轮廓模型, 它能够清晰地定义感兴趣目标模式; 同时构成该模型的Gabor轮廓基元可以通过扰动进行局部的调整以适配图像, 在一定程度上提高了检测算法在目标发生形变或者存在遮挡等情况下的鲁棒性。然后采用交替的sum maps和max maps的计算框架在测试图上扫描出与活动轮廓模型匹配分数最高的区域, 将其定位分割出来; 基于稀疏活动轮廓模型的检测算法在很大程度上依赖于图像尺度, 利用稀疏活动轮廓模型在多个图像尺度上检测物体, 因此可以有效地克服尺度变化大的问题。最后利用视觉皮层模型对分割后的图像进行模式分类进一步确认目标。经过多组实验结果表明, 提出的基于稀疏活动轮廓模型的感兴趣目标检测定位算法, 较好地解决目标在发生部分形变、存在遮挡以及复杂背景下的定位问题。

Because the traditional object localization algorithm based on shape information is difficult to detect the object when the target is viewed from different angle or existing a little deformation. A localization algorithm of object-of-interest (OOI) based on sparse active contour model to solve this problem is presented. Firstly, the active contour model of OOI is learned by shared sketch algorithm, and so the mode of OOI is clearly defined. At the same time, Gabor contour basis that make up of the contour model can be adjusted by local perturbation to fit the image, so it is robust when target occurred a little deformation or the presence of occlusion. Secondly, the framework of sum maps and max maps in turn is computed to obtain the region which matching the active contour template with the highest scores is extracted in the testing image. The detection algorithm based on active contour model largely depends on the image scale, so objects are need to be detected in a number of image scale spaces to overcome this shortcoming. Finally, extracted images are classified and identified according to visual cortex model. The experimental results show that this method can solve the problem of localization when the target is partly changed, occlusive or in complex background.