FISH IMAGE MONITORING SYSTEM FOR DETECTING ACUTE TOXICANTS IN WATER—PROPOSAL OF NEW INDICES AND DETECTION OF NERVE POISONS

ABSTRACT An automatic biological monitoring system for detecting acute toxicities was developed by using image processing techniques. The proposed system aims at the substitution of visual observations by a computer eye, i.e., digital image recognition. It is based on behavioral patterns analyses, in which fish in a test chamber are monitored on line continuously. “The breathing movement index“ proposed previously and other new, advanced indices for more reliable and many-faceted detection techniques are presented on the basis of statistical characteristics derived from two-dimensional distribution analyses for each center coordinate of a school of fish. “The scattering index” is defined to represent the scattering degree of the school. Effectiveness of these indices was demonstrated by using fish image monitoring before and after injection of CN−(0.1 mg.l−1) as well as nerve poisons like fenitrothion (5-10 mg.l−1) and DDT (25 mg.l−1). Detection sensitivities were higher than the 24 hour T Lm (median Tolerance Limit) values, and detection response to these pollutants occurred within 20 minutes after injection.