Cancer is a severe disease with high incidence and mortality rates. Currently, 70% of patients undergo image-guided radiotherapy (IGRT). In IGRT, CT scans, which pose significant risks to patients, delineate cancerous regions, while CBCT scans, with lower radiation doses, target during treatment. However, existing registration methods struggle with aligning CT and CBCT images due to low imaging quality, artifacts, and inconsistent features. This paper presents an image registration framework, FS-OR, which integrates optimizations from both the frequency domain and the spatial domain for CT and CBCT images. Edge features are identified using logarithmic weighted phase congruency edge detection, and then coarse registration is performed using phase correlation techniques. Noise removal for the registered image pair is performed using weighted side-window based gradient guided image filtering. Spatial registration is then accomplished using a feature-based optimization method, utilizing the acquired transformation parameters to enhance the coarse registration results, ensuring a stepwise registration from coarse to fine for CT and CBCT images. Experimental results demonstrate the robust and effective registration of CT and CBCT images.