Traditional urban landscape planning and design often only focus on optimizing local areas or single elements, while neglecting the overall spatial structure and ecosystem of the city, resulting in fragmented urban landscape, lack of unity and coordination. To address this issue, this paper proposes an urban landscape planning and design optimization method based on 3D convolutional neural network image processing. Firstly, through the image processing of urban landscape based on three-dimensional convolutional neural network, the characteristic indexes of urban landscape image are calculated, and the three-dimensional convolutional neural network model is trained. Secondly, the optimization of urban landscape planning and design method is carried out, and the evaluation system of urban landscape planning is constructed through the collection and processing of urban landscape images. Ultimately, a strategy for optimizing urban landscape planning and design is proposed, aiming to achieve enhanced outcomes in urban landscape planning and design. The experimental findings demonstrate that the optimization approach for urban landscape planning and design, utilizing three-dimensional convolutional neural network image processing, outperforms the other two methods in terms of computing urban landscape planning indices. At the same time, this technique shows good performance in the optimization experiment of urban landscape planning and design.