Food is one of the basic needs needed by humans to fulfill the process of developing and growing needs. Serving food can be placed in containers such as plates, bowls, or lunch boxes. Tray box is a type of food lunch box consisting of 4 compartments. Rice, side dishes, vegetables are placed in each compartment so there aren't mix with each other and notice to the nutritional value of the food. An alternative way to know the nutritional content in food is digital image processing technology, by segmenting the image as the first step. In this research, the data used were 31 tray box images (full) consisting of 124 compartment images. The Otsu Thresholding method is used as a method for segmenting food images on a tray box with a color space. Each HSV channel is selected as color feature extraction for the compartment image segmentation process, the average of HSV and RGB are used for the full image segmentation process. The IoU accuracy results for compartmentalized image segmentation on each HSV channel are 0,6058237; 0,9006499, and 0,7726735. The results of IoU accuracy and MSE error for full image segmentation on the HSV average are 0,3069244 and 0,8644671, while the average RGB are 0,2761036 and 0,0267637. Based on the results, the Otsu Thresholding method with color space has good accuracy and provides a small error rate.