In this paper, firstly, we obtain a certain precision image of building roofs in a county with the help of high-resolution remote sensing information technology; then, based on Sobel operator edge detection, threshold segmentation and other image processing techniques, we eliminate non-building information in the image, extract the edge contour, and obtain the roof area of the building, and then estimate the photovoltaic capacity; secondly, we aim to minimise the comprehensive cost of distribution grid, which includes the cost of line modification, transformer expansion, energy storage, and distributed PV investment; finally, we establish a two-layer planning model with the constraints of PV development potential by using quantum particle swarm algorithm. Secondly, with the goal of minimising the comprehensive cost of the distribution network including line reconstruction, transformer expansion, energy storage, distributed PV investment and operation cost, and taking into account the constraints of PV development potential, we establish a two-layer planning model of source-grid-load-storage, and use the quantum particle swarm algorithm to achieve a fast model solution; finally, taking a county distribution network as an example, we verify the practicability of the source-grid-load-storage collaborative planning technology based on the assessment of the county's rooftop PV development potential, which improves the economy and reasonableness of the planning scheme.