光学技术, 2014, 40 (1): 40, 网络出版: 2014-05-23  

一种基于边缘能量图的物体检测方法

An object detection approach based on edge energy image
作者单位
1 燕山大学 信息科学与工程学院, 河北 秦皇岛 066004
2 燕山大学 理学院, 河北 秦皇岛 066004
摘要
物体的变形、边缘中断等因素通常对物体检测的结果影响很大。为了减小这些因素的影响, 提出一种新的基于边缘能量图的物体检测方法。边缘能量图在一定程度上反映了物体的边缘出现的概率, 便于计算并具有较好的鲁棒性。计算出图像的边缘图像, 在此基础上分别计算出模型和测试图像的边缘能量图。基于多分辨率的滑动窗口, 在图像中找出测试图像边缘能量图与模型边缘能量图最相似的区域。基于边缘能量图快速得到目标物体的轮廓。实验结果显示, 该方法使用较快的计算速度得到了相对较高的检测率。
Abstract
Object detection is often affected by object deformation and edge discontinuity. A new object detection approach is proposed to reduce the effects based on edge energy image. Edge energy image gives the possibility of the existence of object’s edge to some extent, which is easy to be calculated and has good robustness. Edge energy images of model and test image are generated from edge pixels extracted from images. Based on multiple resolutions sliding window, the region of the testing edge energy image which is most similar to the model edge energy image is chosen as the detect result. Object’s contour is located rapidly based on edge energy image. The test results confirm that the proposed algorithm achieves fine detection rate with fast speed.

马勤勇, 聂栋栋. 一种基于边缘能量图的物体检测方法[J]. 光学技术, 2014, 40(1): 40. MA Qinyong, NIE Dongdong. An object detection approach based on edge energy image[J]. Optical Technique, 2014, 40(1): 40.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!