Image analysis-based classification of pigmentation patterns in fish: A case study of pseudo-albinism in Senegalese sole

Abstract We present new objective tools for extraction and classification of skin colour and textural features in fish using a Gray Level Co-Occurrence matrix (GLCM) method followed by data dimension reduction procedures. To achieve this, we examined the skin pigmentation patterns in Senegalese sole, Solea senegalensis early juveniles showing several degrees of pigmentation patterns ranging from pseudo-albinosis to normal pigmentation. Four textural descriptors were chosen to extract the textural features of sole skin from the GLCM image (contrast, energy, homogeneity and correlation). Effective classification and discrimination procedures of sole skin textural descriptors were analyzed and compared by means of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). Results indicated that LDA was more suitable for this task and by using it we could easily differentiate between albino and pseudo-albino soles. Therefore, we propose an open source classification system based on image analysis that can be used in studies on fish pigmentation patterns and defects. The description of the work contained herein also suggests how this classification system, together with an appropriately designed mechanical sorting system, might be used in separating abnormal fish during aquaculture production. Statement of relevance This contribution presents new tools for extraction and classification of skin colour and textural features in fish using a Gray Level Co-Occurrence matrix (GLCM) method followed by data dimension reduction procedures. These procedures are of special interest in flatfish hatcheries where pigmentary disorders are common; the use of the image analysis procedures may allow the proper identification of malpigmented fish and the assessment of their quality in terms of pigmentation patterns.