The present disclosure relates to an image processing device and method capable of suppressing a drop in encoding efficiency. One of spatial prediction and phase prediction is set as a prediction method for a pixel to be processed in image data generated by a pixel array in which blocks, each including a plurality of pixels adjacent to each other, are arranged in a predetermined pattern. A prediction value for the pixel to be processed is then derived by applying the prediction method set, and a prediction residual obtained by subtracting the prediction value from each of pixel values in the image data is encoded. The present disclosure can be applied in, for example, an image processing device, an encoding device, a decoding device, an electronic device, an image processing method, a program, or the like.