There are some important factors you've missed that provide clues as to why systems like this don't already exist and invalidate stealth investments.
The short answer is no, because the Koschmieder limit for an F35 sized target is much shorter than that which you propose. A system broadly as you describe will (empirically) detect a target with an apparent size roughly half to three-quarters the size of an F35 (frontal aspect - the only one you should care about for a defence system) from about 6km away (6.3+/- 1.3km). The limiting factors on this are principally optical, and aren't trivially solvable by just adding more pixels.
The longer answer. A 100km distant target has 100km of air between it and you as well, which will emit light (of approximately the same color as the background sky by day), absorb light (from the target) and disort light from the target. Distortions are in the order of multiple arc minutes (for targets that present at about 5 arcminutes in the best case aspect from a distance of 15km). The image you show is the highest contrast and angular resolution case you are likely to find (target at largest aspect and fully silhouetted against the sky), and also probably the most recognisable aspect as well. The visible band offers very poor contrast for this use case, and that is the main reason why IR dominates search and track despite its cost and limitations (because aircraft emit strongly in IR, which helps immensely to push up the SNR and signal to bakcground). Additionally, you must account for loss of information due to the system's own optics.
Systems such as you envisage already exist, but not generally exclusively in the visible band. Fused EO and IRST search and track systems exist, but their practical ranges are far, far shorter than either you envisage or RADAR functions, against fighters. EO search and track has been in use for tracking and gun laying, including AA guns, since the 1970s and predates IR.
There are some important factors you've missed that provide clues as to why systems like this don't already exist and invalidate stealth investments.
The short answer is no, because the Koschmieder limit for an F35 sized target is much shorter than that which you propose. A system broadly as you describe will (empirically) detect a target with an apparent size roughly half to three-quarters the size of an F35 (frontal aspect - the only one you should care about for a defence system) from about 6km away (6.3+/- 1.3km). The limiting factors on this are principally optical, and aren't trivially solvable by just adding more pixels.
The longer answer. A 100km distant target has 100km of air between it and you as well, which will emit light (of approximately the same color as the background sky by day), absorb light (from the target) and disort light from the target. Distortions are in the order of multiple arc minutes (for targets that present at about 5 arcminutes in the best case aspect from a distance of 15km). The image you show is the highest contrast and angular resolution case you are likely to find (target at largest aspect and fully silhouetted against the sky), and also probably the most recognisable aspect as well. The visible band offers very poor contrast for this use case, and that is the main reason why IR dominates search and track despite its cost and limitations (because aircraft emit strongly in IR, which helps immensely to push up the SNR and signal to bakcground). Additionally, you must account for loss of information due to the system's own optics.
Systems such as you envisage already exist, but not generally exclusively in the visible band. Fused EO and IRST search and track systems exist, but their practical ranges are far, far shorter than either you envisage or RADAR functions, against fighters. EO search and track has been in use for tracking and gun laying, including AA guns, since the 1970s and predates IR.