Grape Detection in Vineyards Introduction To Computational and Biological Vision final project Kobi Ruham Eli Izhak

Ideas and goals Vineyards are complex natural environments where accurate detection of fruit can facilitate numerous operations such as targeted spraying and selective picking. Suppose a robot moves along a vineyard aisle and while doing so it takes images of the vines with the camera positioned roughly perpendicularly to the vine. These images might show any of the ground, the vine, and it's supporting structure, the foliage, and grape clusters. The goal of the project is to develop a system that detects grape clusters in vineyard images taken under the conditions specified above (i.e., from within the aisle and toward the vine). By detection it is meant that the program should be able to properly label every pixel that belongs to a grape cluster while labeling every other pixel as non-grape.

Course of action what make the grape different in the vineyards typical image?: grape has shape which is the most similar to cycle but not symmetric. The radius of the cycle is limited. Grape in the image is most likely to be found together with other grapes. The color of the grapes in every image is different, but in the typical images (which was given) it can be found that the color of the grapes is not the dominant color in the image

Course of action In order to filter parts from the given image we examined the following methods: Filter pixel using kmean – remove the biggest segment. Remove pixels which their RGB is out of bounds. Remove x most common gray color in the image. Hough transform for cycle detection: Filter all the edges where the gradient size is very big. The threshold was examined on the number of votes for each cycle. Sum all the votes for nearby cycles and radius and examined threshold on it.

Implementation Step 1 - Pixels filtering: Kmean and then remove the largest segment. Remove all the pixels which their R' G or B is above 200 or below 60. Step 2 - Gray image: change the image to gray scale. Step 3 – Edge detection: gradient (orientation). Step 4 - Hough transform for cycle detection: every edge pixel in the image votes for all the cycles which tangent to this edge point from both sides. The radius of the tangent cycle is bound between 2 and 15.

Implementation Step 5 - Cycles filtering: Step 6 - Cycles Painting Sum all the votes for cycles with difference of one in the radius (i.e. sum all the votes of cycles in location (x,y) and radius 3,4,5). Filter all the cycles which the number of votes (sum of votes) for them is below 10. Keep only the cycle which have at least 35 other cycles around them. Step 6 - Cycles Painting

Results

Conclusions The main method was Hough transform – but the typical image is very detailed and without any help it doesn’t return any good result. In order to use Hough transform so it will return satisfied result parts from the image must be remove. In order to get better result better method for filtering need to be develop or few parameter need to be entered before start detect the grape (distance, common color...). In case no other details are given we find it very difficult to find grapes in all the typical images.