The constant improvements in the technology of semiconductor chips allow increasingly complex functionality to be performed by more and more space and power limited modules. Deep miniaturisation and cost reduction of such modules have positive influence on their application areas. They can be used e.g. inside machine tools for controlling the dimensions of processed elements or a deformation of tool. Modules of this kind need to include as many standard components as possible, especially the processing unit responsible inter alia for processing signal from visual sensor. For this reason in this paper there are reviewed and compared algorithms for detecting edges of objects useful in embedded systems based on standard single chip, Cortex-M4F cores. The processing complexity and the quality of the algorithms outputs are analysed. It is concluded that the required processing times are substantially limiting the use of single chip MCUs to detecting edges on small images.