Most sea-land segmentation algorithms may result in wrong segmentations or lots of isolated areas due to the use of single feature for image processing which makes it difficult for subsequent processing.To solve the problem an improved algorithm for different kinds of land is proposed which combines the features of gray level gradient and texture together.First suitable features are selected to segment the sea and land generally which is followed by morphology processes.Then pixel labeling and texture clustering are used to make the result of sea-land segmentation better by removing the isolated areas.Finally land is shielded by using histogram statistics method.The algorithm proposed makes the detection rate and accuracy higher than 99%.The time spent is only half of the other algorithms.The results show that this algorithm can segment the sea and land and shield the land exactly and effectively and can also make it convenience for subsequent processes.