This contribution introduces two methods based on picture-level geometry transform. Both methods allow only vertical flip geometry transform. In Method 1, a flag is signalled to indicate if the vertical flip geometry transform is applied or not after comparing RD costs in the encoder side. In Method 2, a picture-level edge feature analysis is applied based on the edge types of pixels to determine if the vertical flip of geometry transform is applied or not. Hence, the RD comparing process required in Method 1 is not necessary in Method 2.