It seems like the basic set of stuff that has to be in place for a pramāṇa to occur is: 1) a real object 2) a properly functioning sensory organ 3) a real image of the real object 4) a correctly formed concept based on the real image 5) a prompt to action. The pramāṇa is the most important instrument in prompting action; if everything in this chain is in fact real/correct, telicly efficacious action occurs.
Since the concept is the final member of this chain and only it actually immediately prompts action, it seems like the concept should be the pramāṇa. Dharmakīrti rejects this because, even based on a real image, a concept can be formed incorrectly. Further, one can't form a correct concept unless the image has the causal capacity to produce the concept. So the correctness of the concept depends on the content of the image. This makes perfect sense.
However, why doesn't this same logic extend from the image through the sensory organ to the object? If one doesn't have properly functioning organs, the image won't be correct; if there is no real object, the sensory organ won't sense anything. If the pramāṇa is the thing that MUST be in place for telicly efficacious action to occur, it seems like this would be the object not the image. One can have an image, but unless that image accurately depicts a real object, that image won't lead to a pramāṇa.
Perhaps this is why, in the end, the image IS the object. Since all we can ever know of an object is our experience of it---that is, the image---we don't have any awareness whatsoever of any object other than the image. So it's hard to defend the idea that something we can by definition never have any awareness of is itself an instrumental awareness.
If the image is the object, it's kind of unclear what the sensory organ would be in this case. The image arises as a result of contact between the object and the sensory organ; if the object is already the image, the image is already there, and there seems to be no role for a sensory organ. The sensory organ couldn't be something extrinsic to the image/object because then it would have no role in relating the two. So what is it that allows the image to be the object? Reflexive awareness lets this happen. The capacity for the image and the object to be the same thing is what reflexive awareness provides. I think.
So saying that the image is the pramāṇa is, in the end, saying that reflexive awareness is the pramāṇa because reflexive awareness is a description of what the image of an object in cognition actually is. I think. And I guess reflexive awareness is also a description of how/why the image, the object, the instrument, and the cognition itself are all the same thing. More on this later....
One interesting consequence of this line of thought (if it's in fact correct; that might be a big if ^_^) is that it does not seem possible to justify the claim that the image, not the object, is the pramāṇa unless awareness is of an internal object, not an external object. If the object were other than the image, then the same line of reasoning that claims the image, not the concept, is the pramāṇa would extend all the way to the external object. It seems, then, that while it's easy and convenient to think about the production of a pramāṇa in terms of interaction between an external object and a sensory organ, in the end, Dharmakīrti's account is not compatible with external realism.