Provided is an inter-layer video decoding method including: obtaining prediction mode information of a depth image; generating a prediction block of a current block forming the depth image, based on the obtained prediction mode information; and decoding the depth image by using the prediction block, wherein the obtaining of the prediction mode information includes obtaining a first flag, which indicates whether the depth image allows a method of predicting the depth image by splitting blocks forming the depth image into at least two partitions using a wedgelet as a boundary, and a second flag, which indicates whether the depth image allows a method of predicting the depth image by splitting the blocks forming the depth image into at least two partitions using a contour as a boundary.