【摘要】 研究模板和图象间的有效匹配 ,利用不变矩特征作为检测模板和图象中物体轮廓相似度的测度 ,同时把遗传算法引入图象匹配识别 ,并针对简单遗传算法在应用过程中出现收敛过早和早熟现象的问题 ,采用了一种改进型遗传算法。该算法和不变矩相结合能有效检测出具有平移、旋转和尺度变化的物体 ,该方法可以应用于实际图象匹配和识别中更多还原

【Abstract】 The problem of correlative image matching is studied by means of genetic algorithms and moment invariants. A modified genetic algorithm (MGA) is presented to overcome the slow and premature convergence shortages of simple genetic algorithms (SGA) in applications. There are translation invariants, scale invariants, and rotation invariants in MGA with moment invariants. They can be used in practical correlative image matching and recognition.更多还原