Imaging ladar can offer angle-angle-range-intensity images, which can facilitate target recognition. Imaging ladar has thus attracted much research interests. Imaging processing is a crucial technique for imaging lidar. And image denoising plays a pivotal role in image processing. A successful denoising process can greatly enhance the image quality. Image denoising for coherent ladar intensity images using non-local means (NLM) algorithm is investigated. Both simulated and real images are used. NLM algorithm and homomorphic NLM algorithm are respectively applied to denoise simulated single objective intensity images under different carrier to noise ratios. The results are compared with the denoised results of Lee filter and arithmetic mean filter. Finally, the algorithms are validated by real coherent ladar intensity images. The results show that two-level homomorphic NLM algorithm has the best performance from the aspects of background noise suppression, uniformity on the target and edge preserving.