中文摘要：

宇宙中多数天体在天文图像中呈现点状结构,使得针对天文图像点源探测和提取算法的研究成为热点.提出了一种X射线天文图像点源提取算法.首先,利用阈值分割分离部分背景噪声;然后利用峰值检测的方法获得潜在点源的位置和中心亮度;而后,根据X射线图像光谱的特点,提取点源和背景的光谱特征,利用支持向量机（SVM）进行有监督训练获得分类模型;最后,利用该模型筛除潜在点源中的错误探测.设计实验,应用该算法到NGC 4552星系的X射线天文图像的点源探测.相较于参考算法wavdetect,本算法能够达到相同的误差率（约5%）,但具有更高的处理效率.

英文摘要：

Since most of energy sources in our Universe appear point-like structures,the study of point sources detection method on astronomical images has become significant. In this paper,a point sources detection approach on X-ray astronomical image was proposed. Firstly,a thresholding method was used to separate the background noises. Then,the peak detection method was taken to detect the positions of potential point sources. After that,we extracted spectrum features of point sources and backgrounds,and generated the classification model using the Support Vector Machine. Finally,the correct point sources were got after discarding of spurious detections with the classification model. Our approach was applied to the X-ray image of Galaxy NGC 4552. Compared with ＂wavdetect＂,our approach has the same performance of accuracy with a detection error rate of 5%,but a higher efficiency.