I'm trying to use a ccd linear image sensor (µPD3753) and I'm using the application circuit example in its datsheet, however I'm new at using these type of sensors. Can someone please explain why they're using a hex inverter and not just one (this is needed because the sensor has 2 clocks and they must be inverted in order for the device to work) also why is the pnp transistor needed.
"why they're using a hex inverter and not just one"
...because you need 3 (not just one; their basic function is buffering the signal and just inverting by the way) and they come in packages containing six.
"why is the pnp transistor needed."
...because the CCD output signal is high impedance (cannot provide much current). The transistor circuit is an emitter follower that leaves voltage about the same (just adds about 0.6V but that doesn't matter at all in this application, because the signal will have a more or less arbitray offset anyway because of "dark current") but has low output impedance (can provide much more current).
IC's always define the test setup for input and output parametric results. It says in so many words in the datasheet that this is not mandatory but it is how their IC was tested. It is always a good idea to define source and load impedance for every spec and here they choose what they thought was ideal.
Since the 5V 74HC04 drivers are approximately 50 ohms and the Clock Φ1,Φ2 inputs are 300 pF typ. having a 47 Ω terminates the driver during the transition to minimize ringing from the capacitive load.
The Vout has a Zout spec. of 0.5kΩ typ to 1k max, so having an emitter follower buffers the output signal and with a PNP makes it short circuit proof to gnd only. You can choose whatever impedance you need for your load.
It is consistent with the same attention to detail in the schematic to using 2 caps for supply decoupling and RC decoupling to Vod on IC and emitter follower.