Recently, codebook-based object recognition models have achieved the state-of-art performances in many public object databases. The conventional codebook-based methods treat the whole local descriptor in an image in equal importance. Under such a treatment in a lump, however, all the features from main object as well as from background image have the same importance, and thus the features from background can make undesirable influence on the recognition performance. To solve the problem, we propose a novel object recognition method by combining the conventional coding process and the saliency map. Since the saliency map represents the visual importance of each local area in an image, the proposed method can control each code response using the saliency value. Throughout such process, the proposed method can effectively reduce bad effect corning from low visual saliency regions such as background. Especially, the proposed object recognition method combines the saliency map with the LLC(Local Coordinate Coding), which is the popular coding method. Throughout the experiments on public flower database and Caltech object database, we confirm that proposed method can improve the recognition performances.