The research constitutes the attempt to create new approach to optimization in the field of land use planning. It combines methodologies of remote sensing and landscape ecology, bringing together multi-spectral analysis of digital imagery and analysis of landscape texture. These powerful tools are used to classify and cluster the area of study to the best advantage that can be predicted in developed model. This means that the developed procedure can help to configure or redistribute the area and resources among land use types in a manner that allows maximization of output, which can be received from utilization of the resources. In contrast to the traditional land use assessment and optimization techniques used by USDA and FAO, this methodology does not use linear optimization for individual map unit. When running the optimization, developed model uses the idea of common effort and possibility to bring together all the necessary resources from different map units that can help to achieve the goal of a particular land utilization type. Based on those ideas, the algorithms of semi-lacunarity analysis and edge search were created and combined into one procedure of raster based heuristic land use optimization. Also, the structure of participating data types were designed for the need of proper input data storage and manipulations. The procedure was tested on the soil and terrain data obtained in Wayne National Forest (Ohio). The map of Optimized Land Use became the result of the research and testing. The model helped to exclude ineffective land uses and reassess the land to the effective ones, while keeping their distribution reasonably close to natural patterns of resource distribution.