針對飛機紅外圖像中存在的背景噪聲大、目標幾何結構復雜、圖像對比度低所造成參數活動輪廓模型進行圖像分割的困難，本文利用獲得的各區域灰度信息，進行Chen-Vese 能量函數最小化，進而決定形變點壓力的方向和大小，構造出壓力大小和方向可變的參數活動輪廓模型，來實現飛機紅外圖像分割。改進模型的停止函數不再單一依賴于圖像的局部梯度信息，模型對圖像的復雜邊界和模糊邊界能實現穩定和正確的收斂。
關 鍵 詞: 紅外圖像 Chan-Vese方法參數活動輪廓模型
Abstract:Balloon model based on global information comparison was proposed in the lower contrast and complicated background image. The model proposed by the paper to determine the pressure direction on the basis of global information comparison makes use of the grey information at each area to minimize Chen-Vese energy function so as to determine the direction or size of pressure deformation. In this way, the stopped function of the improved model no longer solely depends on partial gradient information of the image so that the complicated geometrical structure borders and vague borders in the image can still achieve ideal convergence.
Key words: Infrared Image ；Chan-Vese Method ; Parametric Active Contour Model