High Throughput Plant Phenotyping Image Compression and Decompression Algorithm

Abstract

Compressing of images in order to reduce storage space is one of the important tasks in plant phenotyping. With advances in plant phenotyping experiments and adoption of phenotyping activities with genotype process to improve crop yield there will be lot of data generated and hence there is a need for development of compression standards very specific for plant phenotyping images. It is proposed in this study that DTCWT be used to breakdown a picture into several bands in order to reduce the impacts of quantization and increase compression ratio. Each sub band is made up of orientations in six directions, with real and imaginary portions in each direction. As a result of the companding method, all of these orientations are captured and the number of bands is reduced to three high pass bands. The SPIHT method is used to encode the combined low pass and high pass bands, which results in an improved compression ratio. The compressed bit stream is decoded using the SPIHT decoder, and the original picture is rebuilt using DWT, which makes use of DTCWT filters to rebuild the original image from scratch. The computing complexity of the decoder is decreased by 75% when DWT reconstruction is used in place of DTCWT reconstruction, thanks to the use of DWT reconstruction.