In the last years, a scientific and technological field which has developed greatly is the digital image processing. 
This development has been brought about, among other things, by the extraordinary innovations in the production of 
closed-circuits and also by the new computational technology. 

This paper is aimed to analyze the impact that the hardware and software have in a real-time image acquisition system
 used in the study of the predator-prey behavior of sea animals in image processing.

This research stems from the study of the behavior of marine animals, particularly from the analysis of the escape 
trajectories of the shrimp, Litopenaeus vannamei. The events are recorded in video; and later, a researcher checks them
to find interactions between crabs and shrimps. The records of the interactions are videotaped; once the interactions
are detected, each one is evaluated frame by frame to mark the important points and to calculate angles (by hand in the
screen), as well as the speeds (by hand based on a distance measured on the screen and the numbers of frames in an elapsed
time). The initial escape response of the animal is determined by the rotational movements and the kinesthetic energy with
which it moves.  The purpose of this paper is to propose a solution to solve the above automatically so the efficiency of
the researchers increases and the mistakes committed by the manual methods decrease. 


This paper aims fundamentally to design an instrumentation (hardware and software) for the study in real time of the
behavior of different species of sea animals.  The specific experiments performed in this paper were performed on the
response conduct of the shrimp to a natural predator (crab). Nevertheless, the algorithms developed in this paper can be 
adapted easily to the study of other fusiforms and chelae crustaceans .  Due to the time demands in image processing, it 
was necessary to use new methods to decrease the time of image processing. This produces more information about the 
position and movement of the studied animals making possible more accurate predictions of their trajectories. 

A quick review of the basic methods of image processing is introduced; presenting, in a general way, the filters that 
allow the highlighting of characteristics in an image, and emphasizing those that highlight the edges in an image.  The 
edge provides the data necessary to calculate the position vectors. Since, the latter is the longest stage in the image 
processing; a method to shorten this time is one contribution of this paper. 

Improvements in the localization algorithm and the monitoring of the individuals trajectories based on the detection of 
the image edges are proposed. Taking into consideration that the Susan method to detect edges is one of the fastest and 
most reliable, it was compared to the method proposed here. As a result, this stage of the process was performed 4000 
times faster; for due to the improvements, instead of using the general process, a selection of the areas to be analyzed 
is used.
 
To obtain the data, the Channel Link communication protocol was analyzed by implementing it in the Xilinx and Altera FPGAs 
which performs better so the best could be chosen. To perform image processing, a mixed architecture is suggested in which
the DSPs can be integrated to the FPGAs, and even to the encompassed microprocessors, to reduce times. In each phase of 
the project, two solution choices are given in which the best characteristics of each technology is considered so the best
time results are obtained.