PDF(Sci-hub)
Abstract：
Fishes available in the market may be cultured either in fresh or contaminated water bodies. Heavy metals are one of those contaminants which may cause menace to fish health and thereby affect the health of living beings consuming them. The identification of heavy metal residues in fish samples is a challenging task and may require expensive and sophisticated instruments and testing. This paper investigates visual changes which may be used as benchmark for differentiating between fresh water and heavy metal exposed fishes. The proposed method is an automated non-destructive image processing method for identifying visual changes which can be used to differentiate between controlled (untreated) and heavy metals exposed (treated) fishes. The eye of the fish from digital images is considered as focal tissue that was automatically segmented using the Circular Hough Transform and adaptive intensity thresholding. Post segmentation, a potential feature is identified and transformed into mathematical parameters for classification of a fish sample as fresh or heavy metal exposed water fish. The proposed method can identify and translate the potential visual feature for ease of understanding. The accuracy of the proposed method is high, and computation time elapsed indicates the possibility of using such algorithm for real time detection in related field.
摘要：
市场上的鱼类可以在新鲜或受污染的水体中养殖。重金属是可能对鱼类健康造成威胁并因此影响食用它们的生物的健康的那些污染物之一。识别鱼样品中的重金属残留是一项具有挑战性的任务，可能需要昂贵且复杂的仪器和测试。本文研究了视觉变化，这些变化可用作区分淡水和重金属暴露鱼类的基准。所提出的方法是一种自动的无损图像处理方法，用于识别视觉变化，可用于区分受控（未处理）和重金属暴露（处理）的鱼类。来自数字图像的鱼的眼睛被认为是使用圆形霍夫变换和自适应强度阈值自动分割的焦点组织。Postsegmentation,识别潜在特征并将其转化为数学参数，以将鱼样品分类为新鲜或重金属暴露的水鱼。所提出的方法可以识别和翻译潜在的视觉特征，以便于理解。该方法精度较高,和计算时间的流逝表明了在相关领域中使用这种算法进行实时检测的可能性。
