The current estimate of the number of government-run public surveillance cameras in China is 20-30 million. That is a seriously frightening number - that's one camera for every 50 people. This drone will be able to target and permanently disable the optics in these cameras by over-saturating and frying the photo sensor (the CCD) with a high power pulse of light from a diode-based laser.
The deployment of this type of destructive technology is warranted by the unreasonable actions of the communist government that so boldly violate human rights: it is required to restore the right of nearly 1.4 billion people to be secure in their persons from unreasonable searches.
Ensuring the safety of innocent human bystanders will be thoroughly accounted for by implementing multiple safeguards in the design prior to creating a prototype, as mere reflections of high power laser light of various wavelengths may cause permanent blindness. Alternative designs could spray-paint over the lens, disable electronics with high voltage probes, physically sever electronic connections, etc - this might also alleviate the safety concern.
The drone must be fully autonomous and parasitic so it cannot be followed back to its operator. This requires machine vision, target acquisition, and attack follow-through, with no human input. It also requires navigation (altimeter + GPS?) and plenty of memory storage (recording what targets were previously destroyed, future targets, navigation routes, etc). The drone may be networked over wifi or bluetooth or other radio coms channel, with the ground network and/or with other drones. Furthermore, the drone must have a predetermined refueling/recharging station set up clandestinely by the operator that allows for autonomous recharge, or it must be parasitic and steal gasoline or electricity.
Cost should be evaluated relative to the number of cameras destroyed before hardware/software failure or operational compromise occurs. For a robust design, the ideal cost is probably somewhere in the low hundreds of dollars after refining the design. The first prototype might cost a lot more in order to incorporate all the necessary technology using off-the-shelf or oem parts, or anything from McMaster Carr.....