The article reviews recent computational models of focal visual attention, focusing on bottom-up, image-based control mechanisms. It highlights five key trends: the context-dependent perceptual saliency of stimuli, the use of a 'saliency map' to encode stimulus conspicuity, the importance of inhibition-of-return, the interplay between attention and eye movements, and the role of scene understanding and object recognition in selecting attended locations. These insights provide a framework for understanding visual attention from both computational and neurobiological perspectives. The article also discusses the neuronal mechanisms involved in attentional control, including the involvement of early visual processing areas and the dorsal and ventral streams of visual processing. Additionally, it explores the computational principles underlying saliency computation and the integration of bottom-up and top-down cues in attentional selection. The review emphasizes the importance of pre-attentive feature detection and the role of non-classical surround modulation in saliency computation. Finally, it discusses the interaction between attention and object recognition, highlighting the need for more detailed integration of these processes in computational models.The article reviews recent computational models of focal visual attention, focusing on bottom-up, image-based control mechanisms. It highlights five key trends: the context-dependent perceptual saliency of stimuli, the use of a 'saliency map' to encode stimulus conspicuity, the importance of inhibition-of-return, the interplay between attention and eye movements, and the role of scene understanding and object recognition in selecting attended locations. These insights provide a framework for understanding visual attention from both computational and neurobiological perspectives. The article also discusses the neuronal mechanisms involved in attentional control, including the involvement of early visual processing areas and the dorsal and ventral streams of visual processing. Additionally, it explores the computational principles underlying saliency computation and the integration of bottom-up and top-down cues in attentional selection. The review emphasizes the importance of pre-attentive feature detection and the role of non-classical surround modulation in saliency computation. Finally, it discusses the interaction between attention and object recognition, highlighting the need for more detailed integration of these processes in computational models.