Due to the coffee importance to Brazil economy, it is necessary to improve and estimate the area of this cultivation. Thereby, the use of geotechnologys became indispensable to the success of this process. Coffee cultivation are easily mixed with native vegetation in image pixel based classification proccess. The classification of coffee enviroments enables the environmental analysis and regional change analysis, as well as the analysis of many factors that affect those areas, like space time dynamics and environmental impacts. The aim of this study was to classify coffee area with images sattelite high resolution. Those images were segmented with shape priority, samples of the different classes were taken, also using spatial and spectral features caracterization, shape and texture were very important to the class separation, algorithm, was used " nearest neighbor " algorith for classification. In this study were used images from Rapideye sattelite, wich have high resolution spectral and radiometric , was used definiens ecognition software to the segmentation, the sampling and the object classification. Was possible to do the land use map, which was compared to a visual interpreted map and the results of a pixel based classification, getting better results, but still needed a specialist to evaluate the map and to correct it. The Global accuracy and Kappa index to the object classification were 87.536 % and 0,80305 respectively.