随着卫星遥感数据空间分辨率的不断提高，使用传统的基于像元的遥感影像处理方法不仅无法充分利用高分辨率影像中的空间细节信息，还会因为“同物异谱”以及“同谱异物”的现象导致分类结果出现较多的漏分和误分，同时分类结果还会呈现严重的“椒盐噪声”，严重影响了分类精度。应运而生的面向对象分类方法，能有效的抑制上述问题，因此受到了众多学者的关注。纹理信息作为一种重要的影像空间特征信息，在遥感影像分类中有着广泛的应用。众多学者利用纹理信息辅助分类获得了较好的效果。但是目前的研究大多基于传统的面向像元分类，即便是基于面向对象分类的纹理研究也多采用了单一尺度的面向对象分类。并未就纹理信息在多尺度面向对象分类中对分类精度的影响进行深入研究。因此本文针对前人研究，使用唐山市丰南区的IKONOS数据构建了多尺度的面向对象分类体系，并以此研究了纹理信息的添加对于分类精度的影响，得到以下的研究成果：本文通过ESP(Estimation of Scale Parameters)工具和多次试验，确定了研究区域内主要地类最适宜的分割尺度和分割参数，建立了三级的多尺度分割层次（81,47,16），并依此建立了多级的分类体系，体现了多尺度分割在面向对象分类中的优势。在面向对象分类的基础上，提取了8种GLCM（灰度共生矩阵）纹理和3种LSS（局部空间统计）纹理，在SVM（Support Vector Machine）和NN（最邻近）两种分类器下，研究了不同纹理对于总体精度以及各类别PA（用户精度）和UA（制图精度）的影响。实验证明，该实验条件下添加单纹理信息能有效提高总体精度以及大部分类别的PA和UA，在SVM分类器下，纹理信息的添加对总体精度的影响较小，均在1%左右，而在NN分类器下，纹理信息的添加对总体精度的影响较大，Geary's C纹理拥有最佳的总体精度（79.06%），相比单纯使用多光谱的分类结果提升了4.6%的总体精度。选择了部分纹理来研究纹理尺度参数对于分类精度的影响，结果显示尺度参数的变化会对分类结果产生一定的影响，但这种影响会因为面向对象分类本身的机制问题而削弱。提出了一种基于蚁群算法的最优纹理特征组合选择方法，能够在保证较高分类精度的情况下大幅缩减特征维数，可以在未分类的情况下，仅根据样本就可以得到最优的特征组合。得到了两种分类器在实验区内的最优纹理特征组合，并进行了验证。

With the rapid development of spatial resolution in satellite remotely sensed data,thetraditional pixel-based image processing methods encounter bottleneck for the inability to make full use of detail spatial information in high spatial resolution images.In addition, the classification results derived from the pixel-based analysis always have many misclassifications and error classifications, due to the matter of'same material with different spectral' and'same spectral from different materials'. Moreover, there would be serious “salt-and-pepper noise” in the pixel-based classification results, which could drastically affect the classification accuracy. The object-oriented classification method,however, can effectively avoid the abovementioned problems. This method has received extensive attention and been applied in many studies.Texture featureshows important spatial feature information of images, it has been widely used in remotely sensed image classification. Many previous studiesapplied texture feature as supplemental information and got good classification results.But most of the previous studies have focused on the traditional pixel-based classificationmethod, and the texture studies by object-oriented classification with single scale. The influence of texture feature on classification accuracy derived from object-oriented classification with multi scales has not been deeply studied. Based on the previous studies, we chose the Fengnan district in Tangshan city as our study area and generated the multi-scale object-oriented classification system. Through the ESP(Estimation of Scale Parameters) tool and many experiments, we decided the optimal segmentation scale and parameter for the dominant land cover types in the study area. The multi-scale segmentation hierarchy (81, 47, 16) with three levels was built, and the multi-level classification system was established. It showed the advantage of multi-scale segmentation in object-oriented classification.The study extracted 8gray-level co-occurrence matrix (GLCM) texture features and 3 local spatial statistic (LSS) texture features. The classifiers of SVM(support vector machine) and NN(nearest neighbor) were usedto study the influence of different texture features on overall accuracy (OA), producer's accuracy (PA) and user's accuracy (UA). The experiment results showed that the OA and PA and UA of most classified classes could be significantly improved by adding even one texture feature. The added texture information had greater effect on OA of NN classification results, while relatively smaller effect with around 1 % improvement of SVM's OA. Classification results with the Geary's C texture feature had the best OA (79.06%). Compared with classification results generated from multiple spectral bands without texture feature, the adding of Geary's C texture feature improved 4.6% of OA. We chose some texture features to study the influence of texture scale parameter on classification accuracy. The results showed that the change of scale parameter affected classification results, but the influence could be reduced by the self-mechanism of object-oriented classification. This study proposed a method to select the best feature combinations, based on ant colony algorithm. It ensures the great reduce of the feature dimensions, under the condition of high classification accuracy. In addition, the method can obtain the best feature combination from samples without classifying. In this study, we obtained the best texture feature combinations for SVM and NN classifiers in the study area and validated the results.