基于DCT域视觉的场馆监控图像显著区域检测

针对传统图像显著区域的目标运动检测能力不足,提出一种基于DCT域视觉的图像显著区域检测方法.以MCU为基本单位将监控图像分割成图块,利用DCT系数对颜色、亮度、纹理等特征向量进行提取计算得到全局对比度,运用最大类间方差法计算阈值将图像分割,获得视觉显著区域.同时,以全局背景判断图像中的注视区范围、注视点位置,计算注视点...

Salient Area Detection of Venue Monitoring Image Based on DCT Domain Vision

Currently,the traditional detection ability of target in salient region of image is poor.Therefore,this paper proposes a DCT domain vision based salient region detection method.The monitoring image was divided into blocks based on MCU.Through the utilization of DCT coefficients,the color,brightness,texture and other feature vectors were extracted and calculated to get the global contrast.Based on the maximum inter class variance method,the threshold was calculated to segment the image,thus the visual saliency region was obtained.Meanwhile,according to the global background,the range and position of fixation point in the image were determined,and the strongest position in the fixation point was calculated.Eventually,the region detection of salient image was achieved.The simulation results show that the method has high detection efficiency and real-time.