• Projects Overview
• Rocket Aerial Camera
• Wine Imaging Device
• Stereoscopic Imaging Device
• Digital Streak Camera
• Time-Lapse Garden Sun Tracking

Wine Imaging Device

The first aspect of wine tasting is to view the wine in the glass at an angle to observe the complexion and color characteristics of the wine. Each grape which comprises a wine has its own unique properties which will alter how the wine looks. Other factors also cause variations and differences in the appearance, such as age and growing environment. With notable variations it seems possible to use an imaging system to determine the type of wine, location, and vintage.

Prototype Design and Fabrication

From a design stand point there were three major components which were essential to build a working prototype: a camera, a light source, and a wine vessel. These three parts were built into a complete system which allowed for the necessary images to be captured and image processing conducted.

Camera
The camera used for this project was the Microsoft VX 6000 webcam, with a native resolution of 1280 H x 1024 V pixels. This was chosen based upon the cost to resolution ratio, which provided the necessary image quality for the lowest price. It was purchased for $49.99 during a sale; however other models would also be adequate to achieve similar results. The camera was connected to the computer via USB and controlled by Microsoft LifeCam software. This required the images to be saved to a computer then loaded into the analysis software.

Light Source
Creating a light source was the most crucial task of the project because it ensured the color accuracy and highest quality of the resulting images. To achieve the best image possible a uniform and intense light source had to be created. Since all light sources radiate from a single point outward they can not simply be placed in behind the subject being imaged in this situation. This constraint required the light to be uniform across the whole testing region, which could be achieved in two different methods: reflection lighting or diffuse material.

For this project reflection lighting was utilized because it provided more uniform light and the fabrication was less time intensive. The design included creating a box which had a hole for uniformly backlighting the wine vessel. The inside was painted flat white which reflected the maximum amount of light inside the box and caused the light to be more intense.

Inside the box two arrays of 7 Bright White LEDs were aimed accordingly to provide a light source without hot spots. This was be checked by using the webcam to see if the light radiating through the vessel was evenly illuminated. The more LEDs used the brighter the test light would become; the brighter light source would allow for more exposure control and would produce better results.

Wine Vessel
The original design of the wine vessel was a curved glass shape which mimicked the construction of a wine glass. This might prove to be a viable option in the future however it was far too work intensive for a prototype. The design was changed to a linear step concept which mimicked photographic sensitometry test equipment. It was comprised of several panes of plastic sheeting cut so when overlapped there would be an incrementing range of depths. These were then bolted together so that the assembled part would hold liquid while still able to be deconstructed for cleaning.

Image Processing

The program language used to create the Wine Imaging Program was IDL (Interactive Data Language), it was chosen because its primary application is imaging. To determine the type of wine that was being imaged a series of logic based classification statements were necessary. This was achieved when the color properties of the wine returned a true statement causing the program to output the type of wine as a result. The basic functionality which served as a proof of concept was having the imaging system determine whether the wine imaged was a red, rose, or white.

Due to the lower cost and lower camera used in the project the Red, Green, and Blue color space was used. This allowed for easier image processing using the native 8-bit camera resolution which provides 256 lightness values for each color channel. Equations 1,2 and 3 have the red, green, and blue inputs from an image taken of the wine.

Red Wine = (Red Channel < 210, Green Channel < 20, Blue Channel < 40)      (1)

Rose Wine = (210 < Red Channel, 80 < Green Channel < 230, 80 < Blue Channel)      (2)

White Wine = (230 < Red Channel, 230 < Green Channel, Blue Channel < 160)      (3)

The user selects a region by clicking in the initial location and dragging the mouse to the final location and releasing, which causes an array to be created from the values enclosed. This is then averaged to more precisely determine the color of the wine in regards to Red, Green, and Blue values.

Depending on the color of the sample being measured one of these statements shown in equations 1,2 and 3 would be true thus returning one of the screen shots shown featuring the different wine colors.

Initial Testing Results

The fabrication of the prototype was positive and produced a tangible and working device. Specifically, modifications will have to be made to the light source and how its mounted and the wine vessel need to be taken back to the drawing board.

In addition the software achieved a level of functionality which served as a proof of concept of what the project could advance into. It is still very much in the development process and will have to be worked on more to become a viable image processing solution.

Future Work

Improved Wine Vessel
The wine vessel was created using lexan sheeting and nylon hardware. The weakness of the hardware caused the first version of the vessel to be unable to hold liquid without leaking. The next test for this components would be testing with steel hardware and steel brackets so more pressure can be achieved in an effort to retain the liquid without leaks. A future design may step away from a pressure compressed materials to a solid machined vessel.

Multi Level Color Analysis
In the future, use of the entire step target would allow for each color across the different depths to be acquired and analyzed. This would allow for more test points which would result in more separation in color between the ifferent wines allow for better image processing results.

Wine Database
For best results in the future, several different brands would have to be imaged for each type of wine from several vintages and regions. These colormetric tests would provide a basis of the different wine appearances and colors which would allow for test samples to be compared to the database. This would allow for the type of wine, region, and vintage to possibly be calculated.

Sparkling vs. Still
Using image recognition software the bubbles of sparkling wine and champagne could be detected thus allowing for more wines to be analyzed.

Color Calibration
After an appropriate light source is found it needs to be calibrated so the color properties can be neutralized. This would result in an increase in accuracy of the wine imaging system from the resulting images.