Well, first off, its highly doubtful there is a color negative "filter" that would invert the negative in such a way that it came out color positive on the sensor. Color negative film is converted to a color positive print in a multi-step process, ultimately using the film itself as a color filter when enlarging it onto photographic paper with the correct colors of light.
If you had a camera capable of multiple exposures and the proper blending modes (like the Canon 1D X or 5D III), you might be able to simulate the process by putting the color negative film on a backlight that could emit the proper colors, and multi-expose the same frame multiple times under differing backlight. I can't offer much insight as to how well that would work. You could also try the same trick, only take three separate frames and blend them during post processing to achieve roughly the same effect (although with the added benefit that you could use RAW.) You could actually color-correct each CMY color layer that way, before blending into a digital color negative, before inverting into a color positive. You would still need to use a proper color negative inversion algorithm to turn the blended result into a color positive digital image, though.
Just out of curiosity, are you intent on using a DSLR rather than a scanner with proper negative strip scanning support?
I own a relatively cheap Canon CanoScan 8800F flatbed scanner. At the time I bought it, I think I spent about $150. I believe the current replacement for it, the CanoScan 9000F, is around the same price (and its faster to boot). It came with a dual film strip 12-slot holder for negatives and transparencies and medium format 4-slot transparency holder. This relatively cheap flatbed scanner supported scanning film at a fairly incredible 4800x9600dpi. Scanning film at that resolution was relatively slow vs. scanning a book page or other single letter-sized sheet at a much lower resolution (say 300dpi), however since it can scan 12 35mm frames at once in extremely high resolution, it was relatively fast overall (and probably much faster than photographing one color negative at a time.)
Its doubtful you would be able to achieve that kind of resolution with even a high resolution DSLR. The Canon 7D, the new Nikon D3200, and even the Nikon D800 all offer lower resolution than that on the long edge (while the D800 offers slightly more along the short edge.) You would also have noise, vignetting, distortion, diffraction and other aberrations to contend with when using a DSLR, where as that won't generally be a problem with a negative flatbed scanner. Once you scan color negatives into a computer as negatives, you simply need to use a color negative inversion tool in something like Photoshop, Gimp or similar tools to convert them into color positives. At that point, you should be able to correct for color and exposure normally with curves, levels, etc. You won't have as much correction flexibility as you might with a RAW image format, but the original image quality should be higher than if you photograph them with a lens an camera (unless you have a truly stellar lens that exhibits low vignetting and low aberration, and is capable of transferring a maximum amount of resolution to a very high resolution sensor.)
(One tip I have for scanning film is to make sure you have a nice soft brush, like a camel hair art brush, on hand for dusting off the film before and after putting it into the scanner. You'll want to dust off the scanner bed itself, dust off the film and film holder (under side, that faces down into the scanner), as well as dust off the top of the film once it is placed in the scanner. At 9600dpi, the scanner picks up every single detail, and even small bits of dust, hair, etc. can be picked up, create shadows on the film, etc. You want to get rid of as much of it as possible before scanning.)
Balanced Color Negative Inversion: Simple Form
With Adobe Photoshop, or a similar tool with the same functionality, inverting a color negative to a color positive that can be properly color corrected can be a fairly strait forward process. The key tools needed for a proper color negative inversion with this method is a precise levels tool that can be used to set white and black points, an invert command, and color balance adjustment. Lets start with a simple negative (I have none of my own, so I found this one with a Bing search):
If we simply invert this negative without proper preparation, the color positive inversion will usually have a fairly strong cyan cast to it:
The significant orange color cast is due to the classic orange film substrate used in color negative film. Normally, it is part of the color filtration process that allows properly balanced colored light to be projected through the negative in an enlarger onto photographic paper to produce a proper photographic color positive print. All of that mechanical process is meticulously designed and works in a very specific way. A color negative scanned into a computer is unlikely to exactly match that process, so some additional correction needs to be done before inversion to compensate, and some post-inversion color balance adjustments are often needed to fully color correct.
The first step is to adjust the white and black points of the original color negative image, before inversion. The simplest approach can be done by adding a Levels adjustment layer in Photoshop, and simply clicking the "Auto" button:
This will pre-tune the image for proper inversion within photoshop itself. Once levels have been adjusted, you can now invert. Just add an Invert adjustment layer above the Levels adjustment layer. Depending on the film used or the state of the negatives, there may still be some color cast in the resulting image. In this case, there is still a slight cyan cast, however the results are far better than our original attempt at inversion:
The still-present slight cool color cast could be the result of a variety of lighting or development conditions, and some of it may be due to the simplistic nature of this approach to color negative to color positive inversion. Now that the extreme cyan cast of the unprepared direct inversion is gone, color correction can now be done with the standard color balance tools. There is quite a bit of cyan/blue in this photo, so I've added a Color Balance adjustment layer with the settings C/R +100, M/G -10, Y/B -20:
Almost perfect, if a touch on the cooler side. If you want the results to be even warmer, you can add another Color Balance adjustment and tune more to the reds/yellows. This final sample has an additional C/R +50 adjustment:
Looks quite perfect now. Further tuning could be done to taste (I've added a bit of magenta tint to eliminate what looks like a very slight green cast on my screen, however that may simply be a consequence of my calibration, and it may look different on other peoples systems.)
This approach is very simplistic, basically an Auto Levels and Inversion to get you into an easily color-correctable state. More advanced and precise techniques can be achieved via the Curves tool and RGB curve adjustments alone, resulting in perfect inversion with a single adjustment layer. It can be more time consuming, but the results can be tweaked and perfected quite a bit more than with the simplistic approach.
Balanced Color Negative Inversion: Curves Form
To invert a negative with curves, simply add a Curves adjustment layer over your negative. You will want to use RGB curve editing. There are four base curves in RGB editing, your R, G, and B channels as well as a "black" curve, which can be used to adjust contrast (well call it the contrast curve from now on). You use the curves to both invert as well as tune the color of a negative base. To invert, for each color channel (except the contrast curve), drag the corner black point from the bottom to the top along its edge, and the white point from the top to the bottom along its edge. This will invert the negative much like the direct untuned inversion from the simpler approach:
Once inverted, you'll need to tweak the results. Generally speaking, you will want to adjust the white and/or black point of each channel by dragging them horizontally towards the point where that channel's levels fall off to zero. You will frequently only need to change either the white or black point for any given channel, its rarer that you will need to adjust both white and black points. This will begin to eliminate the extreme cyan color cast. You may want to fine-tune the points for each channel by either placing it before the first blips of tones begin, or even after tones begin, or any other extreme...to artistic intent. An initial white point adjustment will cause your curves to look something like this:
Once white and black points have been set, you will need to tweak the bow of each RGB curve to balance color and style the photo as you see fit. This can be a time-consuming process, and depending on the range of tones for each color channel of the original photo, the results may differ. Its useful to give each curve a toe, where the tones peter off near the black point, as that helps with contrast and can help eliminate potentially unsightly artifacts (especially if you set your black point within the tonal range for a given channel.)
The above results look great, and from a color balance standpoint, seem to look better than the final results of the simpler method (which still has a bit of a green cast to it.) You still have the option of tuning contrast independently of each color channel at this point. A little tweaking of that black "contrast" curve can further tune an inverted negative photo:
