<p>Sketch face recognition (SFR) has advanced significantly as a result of these techniques. The hierarchical architecture of the DL techniques helps sketch deep face recognition system learn discriminative face representation. Sketch face recognition faces major challenges with existing techniques, as they are highly sensitive to noise, rely on consistent image quality, and struggle with variations in sketching styles. To overcome this complication, Dynamically Stabilized Recurrent Neural Network Optimized with Wolf Bird Optimization for Face Sketch Recognition (DSRNN–WBO–FSR) is proposed in this paper. Here, the input images are collected from faces catches dataset. Then, the input images are pre-processed with the help of Unscented Trainable Kalman Filter (UTKF) for remove noise, normalize, detect image and enhance image quality from the input image. Next, the pre-processed images are fed to Iterative Matching Synchro squeezing Transform (IMSST) to extract the global features, like colour, histogram, edges, texture. Then the extracted features are supplied into the Dynamically Stabilized Recurrent Neural Network (DSRNN) for sketch face recognition. Finally, the Wolf Bird Optimization Algorithm (WBO) is considered to optimize the parameters of DSRNN. The DSRNN–WBO–FSR is implemented and its effectiveness is evaluated under some performance metrics. The DSRNN–WBO–FSR achieves 29.8%, 21.2%, 18.9% better accuracy, 24.7%, 32.5%, 29.6% better f1-score, 25.8%, 28.5%, 21.6% better sensitivity when compared with the existing approaches: sketch face recognition model using local–global adapter (SFR–LGA–SGCN), Multi branch method for cross-age sketch face recognition (MBA–SFR–CASKNET) and sketch face recognition under light semantic Transformer network (SFR–LWVT–ResNet18) respectively.</p>