To address the difficulty of image analysis in a complex environment, an image segmentation and detection method of an HSV spatial model was studied. First, a UAV was used to collect images. Second, Gaussian convolution was used to detect the gradient of the cracked image after the photovoltaic(PV) panel area was extracted. Finally, morphological image processing and the HSV spatial model were applied to extract the occlusion, and the ratio of the minimum external rectangle area to the PV panel area was calculated. This method can effectively segment and detect PV images in complex backgrounds and provides certain innovative and practical value.