论文部分内容阅读
图像中带状目标在行列上具有规则的几何特性,其边缘映射图在笛卡儿坐标系下的正交投影具有显著性特征。由此提出了一种基于正交投影特性的带状目标检测方法。基于Hough变换的思想,引出一种投影变换,利用投影宽度谱的局部峰值定位边缘映射图中的带状目标。为减小噪声点影响,给出一种距离约束准则下的序贯最小二乘直线拟合方法。对两组图像数据进行了算法验证,实验结果表明,算法识别精度高、实时性好,具有很强的鲁棒性。
In the image, the banded objects have regular geometric features in the ranks, and the orthographic projection of the edge map in Cartesian coordinates has significant features. Therefore, a strip-shaped target detection method based on orthogonal projection is proposed. Based on the idea of Hough transform, a projection transformation is derived, which uses the local peaks of the projection width spectrum to locate the banded targets in the edge map. In order to reduce the influence of noise point, a least square linear fitting method based on distance constraint criterion is given. The algorithm verifies the two sets of image data. The experimental results show that the algorithm has high recognition accuracy, good real-time performance and strong robustness.