Video Camera - Science topic

Thank you Cristina, your work is interesting and helpful.

Mingzi Xu There are various types of video cameras that may be used to follow small animals over broad areas, each with its own set of benefits and drawbacks. Here are some alternatives to consider:

1. High-speed cameras are intended to catch fast-moving objects and may record at high frame rates of up to several thousand frames per second. They can take precise photographs of small creatures at high speeds and quality, but they are highly costly and may require specialist data analysis equipment.

2. High-resolution cameras: These cameras contain a large number of pixels and can capture photos with great detail. They can be used to track small animals across huge distances, although they may not be accurate.

3. Wide-angle cameras have a huge field of vision and may capture a big area with a single photo. They can be useful for following tiny animals across broad areas, although they may be less effective at getting detailed photographs of the animals.

4. Tracking cameras: These cameras are built primarily to track moving objects. They can be used to follow tiny animals across broad areas, however they may be less effective at collecting detailed photographs of the creatures.

It's crucial to remember that, depending on your project's unique requirements and budget, you may need to mix different cameras to attain the necessary results.

The Phantom VEO 640S camera is an option for your application. It's a high-speed camera with a 640x480 resolution and a frame rate of 4,800 fps, as well as a wide angle field of vision. It's worth mentioning that this camera isn't cheap, so you might want to look at other possibilities.

Finally, the finest camera for your project will be determined by your individual needs and budget. It is advised that you examine the specs of several cameras to your personal demands and budget. You may also wish to check with specialists in the subject for additional particular advice.

You are right, doing anything parametric in NMDS-space is a silly idea. PERMANOVA won't help, it's asking a similar question to the ANOSIM you've done. To relate community composition to drivers (e.g. explanatory variables, singly or in combination) there are many methods available. If I was using PERMANOVA I'd probably go for DSTLM (distance-based linear modelling) though my preference would be to stick with the non-parametric approach underpinning ANOSIM and use a method such as BIOENV, which searches for subsets of explanatory variables that give the best match with the biotic similarity matrix.

This would be a great question to post in our new free medical imaging question and answer forum ( www.imagingQA.com ), there are a number of MRI experts in the community. If useful, please feel free to open a new topic at the link below :

I agree with Aparna-Murthy CCTVs is the best option.

Read this article:

How to choose the right statistical test?

By : Barun K Nayak and Avijit Hazra

Some studies have shown that light traps equipped with halogen lamps have outperformed traps equipped with neon lights in terms of their attraction to corn stalk borer butterflies

1. Look at the devices with which you can track the amount of energy consumed. And bring these indications closer to the indications of the optimal process. 2. To reduce the effect of noise on the environment, which includes the personnel operating the oven, it is necessary to equip them with effective protective equipment (goggles, earplugs).

Francis,

maybe I need to explain better.

One camera produces in one night 40 clips of 3 minutes, total something like 8GB. Most of the stuff there is nothing. Sometimes the camera gets activated but the animal passed before the recording started. Sometimes the animal is only 10 seconds to be seen, but runtime is set to 3 minutes. Every clip has date and time stamp. I do not loose any information if I remove all scenes without target animal. Time elapsed between shots can be determined easily. After reduction there is maybe 10 minutes of mp4 left, size 40 MB. That´s a reasonable size to store and to analyze behaviour. And if one would want to go later through the videos, no need to search 2 hours of mostly empty raw material, the compressed information is available. To do this summary manually takes minimum 2 hours. Boring and total waste of time. Some nights I do have set out 4 or more cameras. This means, somebody has to sit all day long and stare into a screen. I got a lady who does this first step for me. The interesting part starts when I get the edited movie. The direction of my research is the interaction between animal and trap. Trap shyness, trap resistance, humane killing, trip force, trigger function, time between first contact and catch etc. Things would be much cheaper for my customers if I could do this automatically.

Sportscode, Dartfish.

Im not sure whats your aim from this reading but why you don't try pressure sensors

Hi Luis,

I think it really depends on what you are exactly doing. Maybe it is worth to think about to use a Raspberry Pi with a camera module. We used such a setup to film toads for 24 h and it worked quite good. Some of the available camera modules are able to film in the infrared spectrum, so you could set up some IR lights to illuminate the area during the night without irritating you experimental animals.

Best, Christoph

run a randomized controlled trial!

I agree with Oscar Páez-Rondón and Laith Al-Ani answers.

The Cons, just quoting.

Our lab uses small HD camcorders setup on tripods, with the camera wrapped in a homemade rain-resistant cover (with a hole on the bottom where the tripod connects to the camera). So, its like the camera is wrapped in a tube. The camera is then connected to a powerbank with USB chord, and the powerbank is inside of dry bag. This setup lasts us 12 hours of HD footage no problem. Last year we had 20 cameras out filming almost every day for 4 months in humid rainforest and they all survived. The setup is small and mobile and I would often carry 4 to setup at multiple nests in the morning. Here is a picture of a setup (the white rope is not part of it)...

If you want proof of their ability to film in a downpour, check out this clip...

(the quality doesn't look as good when viewed on dropbox, but you get the idea)

Hi Danielle,

There has been considerable work that has been done to measure anxiety especially using facial features. This is attributable to the fact that the facial muscles are intimately connected with the task of speaking, performing or simply conveying emotions. The main manifestations of anxiety on the human face involve the eyes (gaze distribution, blinking rate, pupil size variation), the mouth (mouth activity, lip deformations) , the cheeks, as well as the behavior of the head as a whole (head movements, head velocity). Additional facial signs related to anxiety may include a strained face, facial pallor and eyelid twitching. I would recommend you to go through these two articles I am attaching underneath to get a better understanding of what has already been done. I think an AI which can look for, then grade these features can potentially be used in a lot of scenarios. I have attached also a paper that discusses a depression scale but uses a very interesting way to characterize individual facial structure that can be applied to the anxiety scale as well. I hope these help you.

Regards,

Sahil

this paper may help you:

Read the source:

We either use a go pro for low resolution recordings or Dalsa Genie HM640 for 250Hz high resolution, but I think that could be too high for you!

We use streampix to gather and export videos.

For the distance use the disparity calculation.

Daniel Cremers does some cool stuff there.

Hi

Have you considered CCTV cameras? Standard security CCTV cameras have infra red LEDs for night filming. An HD-TVI (or Turbo) CCTV camera plus an suitable recorder will allow scheduling, should you need it, and the ability to film with motion detection or continuously. Hikvision have a good range and can be relatively cheap. You can usually record 4 cameras at a time, should you want to. If mains is available, they can be viewed directly as well.

The cameras give a good quality HD image.

The cameras use 12 volts, and can either be used with mains power (with a suitable plug adapter,) or with 12 volt batteries. Lithium batteries from Tracer power are the best. For CCTV equipment contact Handykam (Cornwall UK) as they are experienced working with researchers.

Interesting...

Hi Klára

You are very right that with IR cameras individual recognition is painful and you may need to discard many photographs. If you are still using Reconyx, please try the following setting:

Sensitivity: High

Night Mode: Fast shutter

We were using Reconyx Hyperfire HC500 for tigers, but as I mentioned many pictures will be blurred and difficult to identify.

Considering your requirements, you can try Cuddeback C or Ambush models (white flash). The shutter lag is very less (0.25 sec), and they have a considerable flash range (40-45 feet). But in higher temperatures (>50 degrees C) battery might get exhausted faster than usual (4-5 days depending on the number of night photographs).

Cheers,

Ayan

Dear Bruno,

regardless the explanation, I think an easy way to reduce the abberation before or after image acquisition would be optical or digital color filtering. Thus, you could place a color filter in front of the camera, or use only one color channel (for example the red channel) of the digital photo.

Kind regards,

Sascha Grusche

I agree with you Dra. Aparna Sathya Murthy.

Regards!!

Hi Jennifer,

If you set up cages on racks, with the side of the cage facing out, you can use any decent HD camera on a tripod to record 6 cages at a time. This will not work on racks with automatic water systems. Going in to turn on the camera will not be disruptive for more than a few minutes, if at all, or you could get cameras that have timers. These cages are going to be in the observation room for all of lactation, without any moving of the cages for the observations or sharing the room with other investigators? The challenge with doing undisturbed observations is tremendous variability in lactation patterns, and therefore maternal care, between litters. You either need to watch many hours of undisturbed video (similar to the Meaney work), or use a short separation to standardize the lactation pattern a bit. A comparison with maternal-pup separation would be a cardiovascular stress test. You could record days of unstressed data from a patient and not see the major pathology that is present during a stress test. For maternal rodents, you probably don't want to introduce a stressor, but animals are handled during cage changes, and a dam leaving the litter for 30-60 minutes is common and does not have the adverse effects of a 3 hr. or longer separation. Trying to use a observation in between observation large durations of undisturbed data and the use of pup separation makes it difficult to see differences. A related issue is recording hours of video and assessing only behavioral frequency through scan sampling, not behavioral duration. Increasing frequency is not always associated with increased duration, and assuming this can lead to inaccurate conclusions. For example, a dam moving her nest might produce increased frequencies of maternal behaviors which result in decreased durations of grooming and nursing. A major challenge with maternal behavior is that you can't use automated assessment for things like pup grooming and nursing, so doing undisturbed recordings can result in a need to observe many hours of video even in a moderately sized study. We have compared results from undisturbed and pup separation maternal care in our own facility, and the durations are very similar with 30-60 minute observations.

I have been using Anymaze software for tracking mice and rat  for three years and it is really a fantastic software for EPM, NOR, FST, fear conditioning,open field, y maze,social interaction,water maze. It is more accurate than manual scoring for freezing , immobility, object exploitation within two centimeters,mobility,time in open arm. Apparatus and Protocol set up is very very important. Once you set up everything, just put in your animal,then you will have the result. I can share my protocol if you want.

A brief research into 'lifelogging' tech gave me these results:

 

(the last one being still in development)

You could pick a glass Fresnel lens that is to be used as a pattern, make a reverse copy using molten tin and press virgin Teflon into the copy using moderate pressure, 25 psi and temperature 375 C.

See for Transmissivity

Surface will be a bit self cleaning as well as UV resistant

Marc

Update: Using Debut Video Capture ($35) software combined with the Logitech camera generated a 22 minute video of only 500 MB which Ethovision was able to read. I did not need to convert the file. I saved it in WMV format. Through the Debut program I was able to lock frame rate at 30 frames per second. This set up worked perfectly and cost just over $100.

Hi Shawn

For the camera  suggestion, there are some many good digital camera now. Ideally if you want a fulll hemispherical view, you can look for camera with a 35mm sensors, but if  cropped hemispherical view is okay, most SLR will work. As for lenses, make sure you get a good one and not all are full 180 deg. 

I still use my old Nikon D90 with a Sigma 8mm F/3.5 EX DG lens.

regards

Sylvain

Oliwia,

I am trying to research feeding of berries on frangula alnus using camera traps. I used browning strike force elite trail camera. Got some video of robins at 6ft distance from tree crown but I suspect videos of robin were incidental to other trigger events eg wind triggering branch movement. This has had big implications for my research.

Whatever model you purchase it would be worthwhile testing it at feeding stations that attract robins or other similar sized birds at different distances . Same goes for any species of interest eg if its a mammal try testing with permission at a pet farm that may have it.  I am now testing for detection rate using this method post field work. It would have been better if I had done it before hand but circumstances didn't allow. Sometimes you only know about pitfalls after you spend time and money on equipment and effort.

Also you may like to  search for the following useful papers. the first two are particularly useful I think.

 "Identifying Performance Differences Among Trail Cameras Used to Monitor Forest Mammals" and

"Wildlife camera trapping: a review and recommendations for linking surveys to ecological processes"

and

Management and Analysis of Camera Trap Data: Alternative Approaches

and

An Open Standard for Camera Trap Data

and

Random versus Game Trail-Based Camera

Hi Emmanuel

Small carnivore behaviour is a specialist and challenging application where nobody has done much camera trapping - the big markets are deer hunters in the USA and researcher want still pictures to do population estimates. There are some papers from Australia and New Zealand, and recent papers on interspecific scent marking, in Santa Cruz, Panama and Botswana that I can send you if you have not seen them already.

I have done some small carnivore camera trapping with mixed results. In my experience the most important feature is trigger speed - at ranges where the image of a small carnivore is a sensible size; out to about 5 or 6 m from the camera, a 1 s trigger delay, added to PIR sensitivity being lower at the edges of the detection zone means that videos begin with the animals already leaving the field of view. Fortunately faster triggers (0.5s) are now standard. If you want to record behaviour that goes on for a while you need long videos (with the disadvantage that they chew memory and take along time to review) and a fast recovery between videos or sod's law will determine that the camera will miss the critical bit !.

Among older cameras Bushnell Trophycams struggle with detection when it is hot, and the video quality is marginal. Stealthcams are way too slow.

Reconyx Ultrafires produce very good videos for a very high price, and have a slow response and recovery. Spypoint 11Ds, at abut 1/3 the price of the Reconyx have very fast triggers and recovery, but their video is not as good. Browning have three new models out that look interesting. Their Strike Force Elite HD is mid range with mid range perfomance but seemed to have a dead spot about 5m out.

Putting it diplomatically; the two review sites you link are not as objective as could be wished.The most reliable reviews are those of TrailcamPro's site; http://www.trailcampro.com/, but even they too are in the business of selling trail cameras, so they tend to look on the bright side.

There is a sudden explosion of mid-range camera traps like the on you are considering. It is likely that they are being produced by the same Chinese manufacturers as make the established brands, so in principal there is no reason why they will not work as well as the established brands. What I do notice is that the price on Amazon is about the same as for Bushnell's, Brownings and Spypoints with similar performance, so you are not saving money by going for a new brand.

They quote a very long detection range, but they could have achieved it with a nice cold background and a large target like a human. It is not relevant to recording small carnivores against a hot African background, - I have seen a family of Selous mongoose playing for 5 min or so 4 m from a Bushnell Trophycam (the old model) in the early evening with not a single image captured !! At 20 m a small carnivore will be a tiny speck anyway, especially with the 90 degree field of view. Unless there is autoexposure control, very bright IR lights are actually a drawback for small targets close up because all you get is burned out images. 

Price and performance specifications are not good predictors of actual performance in the field - you really need to try out an actual camera, but that gets very expensive very quickly. It would be encouraging if the suppliers had enough faith in their products to make them available for testing for research applications.

Thnak you Arvind ji

I should correct my comment to say " it cant detect under-water cold-blooded animals."

Thank you very much David. I will use your anaglyph calculator and let you know about the results.

Best regards. Bogdan 

ok if I understood correctly your question. You have difficulty understanding how this can be done.

Well, a microcontroller can handle various communication protocols such as serial, rs232 using pins Rx and Tx. However, rarely micricontrollers have embedded radio modems for wireless transmission and reception of data. In such case then you need an external modem. You will connect your microcontroller to the modem via usb, or rs232 or etherner. The modem will achieve the wireless transmission. The distance is a function of the transmitted power.

Then at the other end, such as a third person, you will need a second modem that will receive the signal, demodulate it and you can receive the data on a pc or laptop or any other hardware of your choosing using supported communication protocols such as rs232, ethernet etc.

I hope this helps you. I am not a supplier of such equipment but if you need help comparing various brands and models then I can help you with that. But you need to determine your requirements. 

hai,

Zoom Q4 Handy HD Digital Video and Audio Recorder - (New) which is available in Amazon for 25000 Indian Rupees. 

this is best for your purpose. 

thank you. 

The Imaging Source has decent and relatively cheap cmount cameras. If you mean by 'çonstant feed' whole day usage, I would not advise a dSLR.

HTH, René

blogs.wcode.org/2015/04/footfall-a-camera-based-people-counting-system-for-under-60/ is a page that show a little system.  if you are interested, the page source code is:  github.com/WatershedArts/Footfall/tree/master/ThreadedFootfall. 

Other software tool is tsps: //www.tsps.cc/

Hi Sanjay,

If you are looking for a relatively simple method, you may use the phase correlation method. That method gives fairly good results for global motion estimation. See https://en.wikipedia.org/wiki/Phase_correlation or a video processing book such as Video Processing and Communications by Y.Wang et al. 

If you are looking for a more accurate and robust method, see, for example, the IEEE TIP paper "Efficient, robust, and fast global motion estimation for video coding".

Gopro camera.

Hello Felix,

Thanks again for your information. That's all right first I will try this. then lets see how I will go.

anyway thanks for your help. hope the same positive response in coming days.

Thank you,

Regards,

Biswas

Hi, 

This is another small good example, May help you.

Hi! I have filmed small lizards (10 cm) between a range of 2 to 5 meters away, with a Sony HDR-Cx290. It is a great camcorder, I can see lizards movements at a long distance. It has a great zoom and resolution in fullHD. Perhaps, if you can film your wasps at a close distance, with a tripod and with Macro, you can see what you want. Another solution could be, film your wasps with a reflex camera and change the objective for small things..

Hope it helps! 

1) Good luck!

2) Trial and error is a reasonable approach

3) The reason to flush mount the LEDs onto the window is to avoid glare. I am assuming that the light must enter through the same window the camera must look through. In that case, if the lamps are any distance behind the window, some of their light will be reflected from the window into the camera lens, causing glare.

If the oil is too dense and dark to allow the light to propagate, you won't see anything.

It won't hurt to try shining the light onto the window with the lights not flush against the glass. And, maybe you won't have a problem with light reflecting off of the window (i.e., glare). But, I suspect you will (and I'd be delighted to be proved wrong!)

Best

Jim Bales

I have done some work in the area when working as a sport psychologist of a professional basketball team that hired foreign  players. However, the word is not in English. At that time I was faced with the same problem you have. My strategy was to develop strategies taking into consideration some of the things I learned from the literature on doctor patient relationship.

At the present time I believe some some textbooks address the issues when talking about group dynamics. Unfortunately, most of the people writing and publishing in sport psychology have very little experience working with professional on "international teams". Most of the time they have done some work with college students or youngsters. High level sport is a very different reality.

I looked at the current offerings and did not see the model that I used.  I will need to double check, but I think the model number was pc50 or pc51.  I believe I also have a couple of the pc206xp, which are smaller, but never used them for a project.  The ones that I did use have a case roughly the size of a sugar cube, and accept interchangeable micro-lenses.  We used these cameras in a helicopter flight study that was conducted both in daylight and at night, and they performed fine.  I have a paper describing this project at http://scanpath.arc.nasa.gov/pub_files/spie05.pdf, it gives the model number of everything except the cameras!?  That was 10 years ago, so it is not completely surprising that the model is no longer available.  At the time they cost around $50.

Another supplier that you might look at is Marshall Electronics.

Thank you Daniel and Peter for your answers. I am new to this networking stuff so it's little bit difficult for me to troubleshoot the problem.

@Daniel, I tried to ping to IP camera but there is no response on terminal so may be it's not yet detected and IP address which I am getting is random one. I tried connecting other device just for checking IP but still it's the same as earlier so I am not sure weather this IP is from device ethernet port or random generated IP.

@Peter - Thanks for details on VLC. I didn't know that. And yes I was mistaken with standart rtp port. After reading on websites I came to know that I have to write socket address after IP so i tried but in a wrong way. I will do some changes in VLC as mentioned and then will give a shot again.

Hi Steve. Many thanks for the advice. I will check out black magic. I know well about rolling shutters, but will be able to pulse the laser.  

Hmmmm, I got this from GoPro:

Hi Sara,

Unfortunately the HERO cameras do not have that capability they also can not shoot very well in very dark conditions. Please let me know if you have any other questions.

Many Thanks,

Cody A.

GoPro Support

ref:_00Do0HJuF._500o02lWyU:ref

if it is in the lab, just use a webcam. you can then use a PC to program time-lapse, etc. an off-the-shelf old generation hd webcam is cheap so you can get many and only use usb ports.

Because thermal camera lenses captures far infrared waves (thermal waves) emitted from an object. Air can only conduct and cannot produce these waves.

Hi Max,

That looks great.

We ran a trial last week with Mobotix camera with a built in motion sensing algorithm. It seems to work well. I hope we'll get another company to give us a demonstration.

Thanks,

Amir

There are numerous adapters that will allow this.  The C-mount a camera standard.  The microscope has a tube diameter that is the necessary fitting parameter. An example that has adapters for most microscopes:

Some adapters increase magnification and some decrease magnification.  The prices have a wide range based on their optical characteristics.  Spot Imaging makes some of the best adapters but they are not the cheapest.

Hi! I agree on Alessio's statement: your application requirements are crucial. I invite you to read our open-access publication about Unmanned Aerial Systems (UAS) on Photogrammetry and Remote Sensing (link attached), so you get an overall feeling on types of platforms, applications and uses in this context.

Hi

You could try a Raspberry Pi PC (£75-100 UK, if you have a spare monitor) with a noIR camera (£15 UK, make sure it's the noIR version) - minimal cost and quite capable - mine was up and running inside a day, using it to view anaesthetised rodent eyes,but it would be equally good at your task. Software is easily downloaded from their web forum.

For example:

Okay sir

Hi Linus

It's true that Reconyx are relatively big. However as others have said they are very good cameras and come in camouflage pattern. I use them in peri-urban areas and secure them on trees with locking cables. I have PC900 Hyperfires which last many months without battery changes. I don't know the minimum range but I have lots of good footage of animals sniffing the cameras. I've used Moultrie (not bad), and Scoutguard (not good) BUT I don't think they are as good as Reconyx. I work in a wet tropical environment and have found out the hard way that if you buy cheap camera traps you get what you pay for and they will be unreliable and deteriorate quickly; not an issue with a Reconyx. Some of the Reconyx models come with true IR flashes. As as Nickolay pointed out, some brands claim to have IR flashes but there may still be a very faint red led visible. My Scoutguards had this type of IR and as a result I have plenty of images of animals jumping back from the flash; again not an issue with my Reconyx cameras.

You will likely need the Image Acquisition Toolbox (http://www.mathworks.co.uk/products/imaq/), a MATLAB complement for camera control (they provide a 30-day trial). Once your camera is detected by this toolbox, you can configure the acquisition parameters (exposure, gain, region of interest...), record images, and store them in arrays for further display or processing.

Hi Valérie,

I work on the aggregation behavior of necrophagous Dipteran larvae. These species do not differ in the red. I used infrared camera in some experiments and also classical camera but I light individuals with a red light.

If you want more details on electrophysiological techniques, you can have some answer in this paper http://www.sciencedirect.com/science/article/pii/S0022191010001885.

Before we can make any recommendation about UV or other filters, it would be helpful to know what you mean by "survey." Do you mean that you want to non-destructively "census" insects at a feeding stations or a light source (or a white sheet with a light behind it)? That is, are you wanting an automatic camera taking shots at regular time intervals? Or do you mean that you are walking transects, using spotlights, and then taking photographs?

Some thoughts:

- If using simple spotlighting to census insects (at night) and then repeating the transects during the day, a good wide-angle lens will no doubt be necessary. For small insects, you may want to build a rig with a reverse-mounted lens with a flash diffuser (on a bracket). (To save money, don't buy a macro lens. Just take your regular 18-55mm--or shorter--lens and turn it around. You need a cheap mount adapter to do this.) You can add to this setup using extension tubes.

- For daytime shots, you may avoid getting close to your subjects altogether. I suggest trying a larger lens (say, a 70-300mm) with a set of extension tubes. You'll be able to stay far enough away to prevent insects from flying but get high enough magnification to capture the insect well.

- Concerning UV lens (@Madan Gautam) : You probably don't need this if you just want to identify the insects. If you do want to get into animal visual modeling (there's a great new literature out there for using standard DSLR cameras for detecting UV), be warned that this is an expensive prospect. The associated equipment can be very expensive. One of the standard ones (Nikon Nikkor UV lens) runs about US$8000. Of course, the lens is only the beginning. Because the camera sensor elements sensitive to UV are also the ones sensitive to other dominant wavelengths (in the human visual range), it is necessary to block out the visual-range wavelengths. So, you need a UV-pass filter. Of course, because you want to discriminate between UV and visual, you also need a filter that cuts out UV and isolates the visual range. These filters are expensive. So, let's hope you're only doing this for survey purposes.

Dear Dr. Niku - have a look on Canon 60D with an excellent resolved and very fast live view with reasonable price relation. T2-Adapter is needed and an USB-2 card for fast live view transfer to your PC. Excellent Canon software package for camera guiding via PC.

best regards

Vitus Stachniss

First using acronyms muddies the request. I don't know what you mean by SW.

I am familiar with a variety of cameras, but without knowing your set-up it is more difficult to make suggestions. The suggested Panasonic may no longer be available and the resolution is not high... If you used a DSLR such as a Canon T4i with a good macro lense and extension tubes to attain the magnification you wish you would have P1080 video which is HD. I am sure you can set up appropriate illumination with the camera on a adjustable stand.

There are also digital ocular cameras that can be use on a trinocular microscope as well as small digital microscope with its own stand that has software to enable video capture.

the benefit of a DSLR is that the sensor is much larger and you can also use the camera for other work as well.

A Canon T4i is probably best value .. about $1000 with a zoom lense

The macro lense ranges from about $600 - $900 depending on focal length

Extension tubes that link to the TTL system are about $275

You would also need a power supply if the time to shoot is fairly long.

Some cameras can also be set up to shoot time lapse through software..

This gives you a very high quality setup with lots of flexibility.

If you go the microscope route, there are lots of options as well, but I think the DSLR route is probably much better

good luck