create an academic review schematic diagram for environmental microplastic analysis, inspired by the structure of the provided reference figure but redesigned with new content. the figure should have a clean white background, soft rounded rectangles, light blue and light orange main areas, thin dark outlines, minimalist scientific line icons, flat 2d illustration style, high readability, and no realistic shadows. overall layout: three horizontal sections from left to right. left section: title "microplastic sample preparation and image acquisition". include three simple icons: water sample collection, filter membrane enrichment, and microscope imaging. show the process as "sample preparation" → "image acquisition" → "input image". middle section: the largest area, divided into two parallel methodological layers. upper layer title: "semi-automated workflow based on traditional image processing". use a pale blue background. include modules in sequence: "image preprocessing", "morphological operations", "threshold segmentation", "edge detection", "manual correction & target screening", "morphometric parameter measurement". add small icons for filtering, binary segmentation, edge outline, manual checking, and particle size measurement. lower layer title: "fully automated recognition based on deep learning". use a pale orange background. include modules in sequence: "dataset construction", "annotation and training set", "model construction", "object detection", "instance segmentation", "automatic detection, segmentation, and quantification". show example model icons such as "yolov8", "mask r-cnn", and "segment anything". right section: title "result comparison and accuracy evaluation". include evaluation indicators: "iou", "dice coefficient", "precision", "recall", "f1-score", "map". use visual icons such as accuracy target, confusion matrix, bar chart, and evaluation meter. add a closed-loop arrow from the right evaluation section back to the middle workflow, labeled "result feedback and iterative optimization". the loop indicates that quantitative results and spectral ground truth can be used for model updating, external validation, and error analysis. at the bottom, add a concise caption: "fig. x. integrated workflow of image screening–coordinate transfer–targeted spectral confirmation for microplastic analysis." use only english labels. keep all text clear, concise, and legible. use soft colors: light blue for traditional processing, light orange for deep learning, light green for result evaluation. the style should be suitable for an environmental science or analytical chemistry review paper.