Single-cell analysis has important implications for understanding the specificity of cells. To analyze the specificity of rare cells in complex blood and biopsy samples, selective lysis of target single cells is pivotal but difficult. Microfluidics, particularly droplet microfluidics, has emerged as a promising tool for singlecell analysis. In this paper, we present a smart droplet microfluidic system that allows for single-cell selective lysis and real-time sorting, aided by the techniques of microinjection and image recognition. A custom program evolved from Python is proposed for recognizing target droplets and single cells, which also coordinates the operation of various parts in a whole microfluidic system. We have systematically investigated the effects of voltage and injection pressure applied to the oil−water interface on droplet microinjection. An efficient and selective droplet injection scheme with image feedback has been demonstrated, with an efficiency increased dramatically from 2.5% to about 100%. Furthermore, we have proven that the cell lysis solution can be selectively injected into target single-cell droplets. Then these droplets are shifted into the sorting area, with an efficiency for single K562 cells reaching up to 73%. The system function is finally explored by introducing complex cell samples, namely, K562 cells and HUVECs, with a success rate of 75.2% in treating K562 cells as targets. This system enables automated single-cell selective lysis without the need for manual handling and sheds new light on the cooperation with other detection techniques for a broad range of single-cell analysis.