Hardware Trojans (HT) become an emerging issue in the integrated circuits (IC) production, which have many malicious effects on the ICs such as functionality changes, overheating problems and leaking information. According to the Trojans’ small size and ICs complex structure, detecting Trojans is a serious challenge to solve. In this paper, we propose a novel approach in order to detect very small Trojans with high accuracy. First, we generate the ideal power heatmap of ISCAS benchmarks with calculating switching activities and the dynamic power of each gate and node of the testing circuits. Second, we employ image processing techniques (similarity structure index and thresholding) to detect differences in the heatmaps which caused by Trojans. With our approach, very small Trojans that consume power less than 0.001 of the total power consumption can be detected and located with 97 percent average accuracy.