Oceanic fronts are regions over the oceans where a significant change in the characteristics of the water masses is observed. Advanced Very High Resolution Radiometer (AVHRR) satellite imagery over the Bay of Bengal shows regions that are populated by frontal structures. Over the Bay of Bengal, some of the strongest gradients in temperature and salinity are observed. In recent years, there has been a tremendous growth in the availability of satellite imagery and the necessity of automated fast detection of the frontal features is needed for services like potential fishing zones over open oceans. In this article, an algorithm to infer oceanic fronts over the Bay of Bengal is described using changepoint analysis. The changepoint algorithm is combined in a novel way with a contextual median filter to detect frontal features in AVHRR imagery. The changepoint analysis is a non-parametric technique that does not put thresholds on the gradients of brightness temperatures of the satellite imagery. In the open oceans, the gradients of temperature and salinity are not sharp and changepoint analysis is found to be a useful complementary technique to the existing front detecting methods when combined with contextual median filters.

海洋前沿是海洋上的区域，在那里观察到水团特征发生显着变化。孟加拉湾上空的高级甚高分辨率辐射计 (AVHRR) 卫星图像显示了由锋面结构组成的区域。在孟加拉湾，观察到一些最强的温度和盐度梯度。近年来，卫星图像的可用性有了巨大的增长，并且需要对正面特征进行自动快速检测，以用于公海潜在捕鱼区等服务。在本文中，使用变化点分析来描述一种用于推断孟加拉湾上空洋锋的算法。变点算法以一种新颖的方式与上下文中值滤波器相结合，以检测 AVHRR 图像中的正面特征。变点分析是一种非参数技术，它不对卫星图像的亮度温度梯度设置阈值。在开阔的海洋中，温度和盐度的梯度并不明显，当与上下文中值滤波器结合使用时，发现变点分析是对现有前沿检测方法的一种有用的补充技术。