CAJ | 학술논문

图像特征点是一个重要局部特征,传统的灰度重心法是重心模型的一种改进,其主要特点是对点坐标进行像素点的灰度加权,但其抗噪声性能较差.对此,文中采用一种改进算法,通过将目标区域划分成内部像素区域和边缘像素区域2部分,并对内部像素灰度进行均值化,从而有效抑制内部像素噪声.同时,利用连通性对图像进行特征标识,提高特征点提取的速度.最后,通过仿真实验验证了改进算法的正确性,并表明改进算法有更好的噪声抑制性能.
도상특정점시일개중요국부특정,전통적회도중심법시중심모형적일충개진,기주요특점시대점좌표진행상소점적회도가권,단기항조성성능교차.대차,문중채용일충개진산법,통과장목표구역화분성내부상소구역화변연상소구역2부분,병대내부상소회도진행균치화,종이유효억제내부상소조성.동시,이용련통성대도상진행특정표식,제고특정점제취적속도.최후,통과방진실험험증료개진산법적정학성,병표명개진산법유경호적조성억제성능.
The feature point image is an important patial feature. In view of the poor antinoise performance of traditional gray weighted centroid algorithm, an improved algorithm was adopted. Object region was partitioned into inner pixel region and edge pixel region, and all grey levels of the inner pixels were averaged to effectively restrain the noise of inner pixels. At the same time, to improve the velocity, the feature by connectivity was marked. Finally, simulation experiments are conducted to prove the validity of the improved algorithm,and indicate that the improved algorithm has better antinoise performance.