In plant phenotyping, there is a demand for high-throughput, non-destructive systems that can accurately analyse various plant traits by measuring features such as plant volume, leaf area, and stem length. Existing vision-based systems either focus on speed using 2D imaging, which is consequently inaccurate, or on accuracy using time-consuming 3D methods. In this paper, we present a computer-vision system for seedling phenotyping that combines best of both approaches by utilizing a fast three-dimensional (3D) reconstruction method. We developed image processing methods for the identification and segmentation of plant organs (stem and leaf) from the 3D plant model. Various measurements of plant features such as plant volume, leaf area, and stem length are estimated based on these plant segments. We evaluate the accuracy of our system by comparing the measurements of our methods with ground truth measurements obtained destructively by hand. The results indicate that the proposed system is very promising.

We propose a new dense local stereo matching framework for gray-level images based on an adaptive local segmentation using a dynamic threshold. We define a new validity domain of the frontoparallel assumption based on the local intensity variations in the 4 neighborhoods of the matching pixel. The preprocessing step smoothes low-textured areas and sharpens texture edges, whereas the postprocessing step detects and recovers occluded and unreliable disparities. The algorithm achieves high stereo reconstruction quality in regions with uniform intensities as well as in textured regions. The algorithm is robust against local radiometrical differences and successfully recovers disparities around the objects edges, disparities of thin objects, and the disparities of the occluded region. Moreover, our algorithm intrinsically prevents errors caused by occlusion to propagate into nonoccluded regions. It has only a small number of parameters. The performance of our algorithm is evaluated on the Middlebury test bed stereo images. It ranks highly on the evaluation list outperforming many local and global stereo algorithms using color images. Among the local algorithms relying on the frontoparallel assumption, our algorithm is the best-ranked algorithm. We also demonstrate that our algorithm is working well on practical examples as for disparity estimation of a tomato seedling and a 3D reconstruction of a face.

We propose a new local algorithm for dense stereo matching of gray images. This algorithm is a hybrid of the pixel based and the window based matching approach; it uses a subset of pixels from the large window for matching. Our algorithm does not suffer from the common pitfalls of the window based matching. It successfully recovers disparities of the thin objects and preserves disparity discontinuities. The only criterion for pixel selection is the intensity difference with the central pixel. The subset contains only pixels which lay within a fixed threshold from the central gray value. As a consequence of the fixed threshold, a low-textured windows will use a larger percentage of pixels for matching, while textured windows can use just a few. In such manner, this approach also reduces the memory consumption. The cost is calculated as the sum of squared differences normalized to the number of the used pixels. The algorithm performance is demonstrated on the test images from the Middlebury stereo evaluation framework.

We introduce a new likelihood function for window-based stereo matching. This likelihood can cope with unknown textures, uncertain gain factors, uncertain offsets, and correlated noise. The method can be finetuned to the uncertainty ranges of the gains and offsets, rather than a full, blunt normalization as in NCC (normalized cross correlation). The likelihood is based on a sound probabilistic model. As such it can be directly used within a probabilistic framework. We demonstrate this by embedding the likelihood in a HMM (hidden Markov model) formulation of the 3D reconstruction problem, and applying this to a test scene. We compare the reconstruction results with the results when the similarity measure is the NCC, and we show that our likelihood fits better within the probabilistic frame for stereo matching than NCC.

No abstract