中文摘要：

为了进一步发掘蚁群算法的应用潜力，提高分类精度，将相关性引入分类规则发掘过程，试图在蚁群算法挖掘规则时既考虑像元的光谱信息，又兼顾邻近像元灰度的空间相关性，提出了一种优化的蚁群算法。算法包括对单个像元的分类规则挖掘和顾及邻域像元相关性的分类规则挖掘，单个像元的分类规则挖掘中，为使信息素缓和增加，避免陷入局部最优解，同时保证算法具有适当的收敛速度，采用自适应方案调整参数。顾及邻域像元相关性的分类规则挖掘中选用了优势类、优势度、类熵和邻域类相关性等4个指标，以反映邻域相关性对分类结果的影响。实验研究发现，顾及邻域蚁群算法的分类结果精度有了较为明显的提高，总体精度提高了3．00％，其优势主要体现在对建设用地、裸地等复杂地物的识别。研究结果表明，顾及邻域蚁群算法能够更准确地提取光谱信息复杂的地物，有效地减弱同物异谱和异物同谱现象的干扰。

英文摘要：

In order to improve the potential applications of ACO （ant colony optimization） and the classification accura- cy, a new ACO algorithm is proposed which use the correla- tion in classification rule excavation, in the process of classi- fication spectral characteristics and spatial correlation are all considered. This algorithm includes mining the individual pixel classification rule and classification rules of the correla- tion between neighborhood pixels. In process of mining individual pixel classification rule, in order to ensure pheromone moderately increasing, and avoid that the algorithm falls into local optimal solution, and ensure satisfied convergence speed, adaptive scheme for parameter is used. In the process of classification rule excavation considering adjacent pixels, 4 parameters are used to reflect the effects of neighborhood pixels on classification results, including dominance classes, degree of dominance, class entropy and correlation. The result shows that： ①the classification accuracy of this new algorithm is obviously improved in comparison to traditional algorithm, and the overall precision improved by 3.00%;② the advantage of our algorithm is high accuracy recognition for building, bare land. The new ACO can be used to get more accurate extract result in classification of complex ter-rain, and it effectively weaken the influences of the same ground feature with different spectral and the same spectral for different ground features.