Automated flower enumeration, a felicitous method developed for the floriculture industries

Floriculture has become one of the vital profitable businesses in Indian agriculture. The important flowers which are internationally traded include lilium, tagetes, rose, tulip, chrysanthemum, carnation, tuberose, crossandra, etc. Estimation of yield at the time of harvesting of flowers is an important aspect in these floriculture businesses which help strategize their marketing. The present study thus focuses on Lilium and Tagetes (marigold), which are well-known cut and loose flowers, respectively. Cut flowers are harvested when buds start showing color while loose flowers are harvested depending upon the varieties when it attains full size. Conventionally, yield estimations are done manually by counting buds and flowers, which is often erroneous and time-consuming. This paper attempts to develop an automated system for counting Lilium buds and Tagetes flowers at the time of its harvesting using digital image processing techniques. The process implicates image acquisition, pre-processing, thresholding, watershed and finally buds/flower counting for yield estimation. The validation of the results has been done by comparing the results obtained through the manual method as well as by automated counting. The entire process was repeated four times with four different photos to judge the robustness of the techniques. The obtained result was 95.61% accurate for Lilium and 96.66% in the case of Tagetes, airing the possibilities of using the approach. The systematic workflow with pros and cons has been discussed in this paper.