A method of the hybrid speckle-pattern compressive computational ghost imaging scheme is proposed. The scheme detects the larger and smaller resolution areas of the object via identifying complex object composed of different resolution scales automatically. The hybrid speckle pattern composed of different sizes of speckles is generated according to the identify areas. The compressive sensing techniques are combined to process the reconstructed image. Theoretical analysis and numerical simulation show that compared with the traditional computational ghost imaging, this scheme overcomes the influence of inappropriate selection of speckle sizes on the quality of reconstructed image, enhances the image contrast-to-noise ratio and visibility significantly, mean square error effectively. The scheme improves the image quality while reducing the sampling time, which further facilitate the practical application of computational ghost imaging.