The invention discloses a depth estimation method based on light-field data distribution, and the method comprises the following steps: S1, adjusting the pixel distribution of image inputted into a light field, and generating a plurality of re-focusing light-field images with different focal lengths; S2, respectively extracting the intensity range of the same macro-pixel aiming at the plurality of re-focusing light-field images, thereby selecting the re-focusing light-field image corresponding to the minimum intensity range, wherein the focal length of the re-focusing light-field image serves as the scene depth of the macro-pixel, the macro-pixel is corresponding to one point in an actual scene, and the intensity range of the macro-pixel is the change range of intensity values of all points in the macro-pixel; and carrying out step S2 repeatedly, and obtaining the scene depths of all macro-pixels. The method provided by the invention can make the most of scene pattern and spatial information collected by a light-field camera like Lytro, and obtains the estimation of scene depths with abundant details, clear characteristics and high accuracy and consistency.