Research On Citrus Detection In A Tree Canopy Using Machine Vision

In order to automate fruit harvest, harvesting robot has been studied for many years in Japan , Europe and America.) It is essential for harvesting robot to detect the fruits and show their positions to the manipulator. In this research, two methods, one based on color indexes and another based on infrared thermal imaging, were put forward and used to detect citrus in a tree. The following are the main research works and conclusions.When extracting color characteristics for citrus detection, fifty-three color images with natural citrus-grove scenes were used to identify red, green and blue components (R, G and B). Target surfaces (citrus, leaves, or branches) were analyzed using the range of interest (ROI) tool. The values of R subtracted by G of the given citrus, leaves, or branches are very similar, but the values of G subtracted by B or R subtracted by B of citrus are different from that of leaves and branches. A rule for segmenting citrus from background was put forward that the pixel belongs to citrus if the value of R subtracted by B larger than the threshold value (T), which was developed by dynamic threshold segmentation method, otherwise the pixel belongs to background. A program was set up using Visual Basic 6.0. The results show that the visible fruits under direct sunlight or backlighting condition were identified with an accuracy of 100%, however, intense background sunlight will cause poor fruit illumination and poor segmentation results.It is widely known that there is thermal radiation difference between citrus and leaves. This research investigated the potential of using infrared thermal imaging to detect citrus in a tree canopy. The infrared thermal images of citrus-tree were acquired under the following conditions.A. The ambient air at 48%RH, 21.7℃, in the afternoon.B. The ambient air at 78.6%RH, 14.9℃, in the evening.C. The ambient air at 76.1%RH, 12.2℃, in the morning.Analyzing the infrared thermal images (condition A), the results indicated that the thermal difference between citrus and leaves was 1.8℃, and the difference between citrus and branches was only 0.6℃. Correspondingly, the gray level difference between citrus and leaves was 95, and the difference between citrus and branches was only 37. Using wavelet-based multi-scale transform to detect infrared target, it is possible to acquire modulus image to segment citrus from background under direct sunlight conditions, whereas it is difficulty to identification because of the confusion of the reflected light from leaves under backlighting. The method is very sensitive to the ambient air condition, such as in condition B or C, the identification is very difficult.